Enriching Animations

Barbara Tversky¹, Julie Heiser¹, Sandra Lozano¹,

Rachel Mackenzie¹, and Julie Morrison²

Department of Psychology

¹Stanford University, ²Bryant University

Problems with Animations

Why do Animations Fail? The stubborn fact is that animations have yet to prove more effective than equivalent static graphics in teaching a wide range of topics (see review of several dozen studies in Tversky, Morrison, and Betrancourt, 2002). What have been taken as successes of animations have typically been confounded. In some studies, animations have been compared to text or static graphics that didn’t contain the same information. In other studies, animations were interactive, but the control conditions were not. Interactivity is known to facilitate learning and understanding, at least in part because it requires active construction and testing of hypotheses by learners. What follows are an analysis of why animations fail and an analysis of a corpus of scientific diagrams. From these, we derive suggestions for creating effective animations.

Caveats. Some qualifications for the claim are in order. The animations that have not proved to be superior have typically shown the operation of a system from beginning to end. Systems include physical systems such as mechanics as well as conceptual systems such as algorithms. They are meant to convey structural or more commonly causal or conceptual content, typically to students in educational settings. Animations are used for many other ends. They are common in computer interfaces to maintain real-time spatial or temporal continuity, and for that, they may be helpful. Animations meant to convey manner or timing of actions may indeed convey those better than the alternatives. Animations such as virtual tours may also be effective for showing what things look like, especially from different perspectives. Animations that are invented and used in exploring phenomena and data, often by experts in pursuit of understanding, may be effective. Creating animations may confer benefits to the creators. Expertise in the specific content and animation may make some animations effective. Our interest is in animations meant to convey more than spatial or temporal continuity, more than what things look like, but less than advanced expertise; our interest is in animations conveying for relative novices, the kinds of information that typical animations in educational settings convey.

What is an Animation and What Could Animations Be? Loosely defined, an animation is a changing graphic display. Although there are numerous ways that graphic displays can change, the typical animation changes continuously in time and shows the operation of a system from start to finish, at the same temporal and spatial grain, from the same temporal, spatial, and conceptual perspective. In general, animations use primarily structural graphic information, without enhancing or highlighting that information. Changing these implicit procedures for animations may make them more effective, just as changing these aspects of static graphics improves static graphics (e. g., Tversky, 2001). There is research encouraging this idea.

Supplementing an animation with a narrative increases its’ effect (Mayer & Anderson, 1991; Mayer & Sims, 1997). Narratives provide several benefits: one is to focus learners on the important parts and the important changes in the animation; a second is to provide an explanation of what is happening. Animations can focus and explain using graphics, supplemented by symbols, including language.

An animation consisting of a sequence of highlighted stills proved to be effective in instilling mental models, even when it did not improve learning. In a series of experiments, students learned the layout of a library or the history and demographic characteristics of an imaginary country. In both cases, the information could be organized in two ways. A series of sequentially presented diagrams—a discretized animation—organized the information in one way for one-third of the students and in the other way for another third; the remaining third served as a control group and studied a single static graphic with no organizational focus. Although the control group that studied a single static graphic remembered as much as the groups that studied animated graphics, the mental models of students in the animated graphic conditions were organized by the animations, as reflected in both speed and accuracy of verifying information compatible or incompatible with the organization imposed by the animated graphic (Betrancourt & Tversky, in press). Another example in which an animation, this time, a continuous one, provided a mental model is in solving the Duncker radiation problem. A hint of arrows moving toward a center proved to be a better hint than static arrows for inducing transfer from a previous analogous problem (Pedone, Hummel, and Holyoak, 2001). In both of these cases, the animations were simple. For teaching the organization of the library or the geography of the country, the animation consisted of showing the same diagram of the library or country with a different cluster highlighted on each showing. For transfer in problem solving, the animation consisted of arrows around a circle, moving to the center of the circle.

Principles of Good Graphics

Congruence and Apprehension. The conclusion that so far, animations have not proved to be superior to informationally equivalent static graphics in teaching structural or conceptual content is met with surprise, dismay, disbelief, even anger. The resistance is understandable. Animations are viewed as more realistic, that is, more like real life. The presumption is that closer to life is better. Animations appear to satisfy the Congruence Principle of good external representations (Tversky, et al, 2002), according to which the structure and content of an external representation should match the structure and content of the desired mental representation. Animations use change in time to convey change in time, a natural, compatible correspondence, often the core of the information to be conveyed. But those trying to figure out what is happening in animations, animations given by the real world as well as animations designed by educators and computer scientists, soon realize that animations fail the Apprehension Principle of good graphics (Tversky, et al., 2002), according to which an external representation should be readily and accurately perceived. Too much happens too fast. Static graphics can be inspected and reinspected. Until stop-gap photography (Muybridge, 1955, 1957), generations of painters portrayed horses galloping incorrectly; the exact pattern of movements of four legs galloping is to complex to be apprehended in real time. Yet, even simple animations, portraying a slowly moving path of a simple geometric figure, are no better than static motion paths in teaching rules of navigation (Morrison and Tversky, in preparation). What’s more, people do not always correctly perceive even simple motion paths; for example, straight paths of moving objects are perceived as more horizontal or vertical than they actually are (Shiffrar & Shepard, 1991; Pani, Jeffres, Shippey & Schwartz, 1996). Selection of the correct motion path from animations of several possible motion paths of a pendulum or falling object does not guarantee correct production of the motion path (Kaiser, Proffitt, Whelan, & Hecht, 1992). Seeing doesn’t guarantee correct perception or correct understanding.

Users often know animations are inadequate even when designers do not. Users often ignore a readily available animation, preferring some other means of learning. Experience may matter. Prior to using an animation of a set of simple navigation rules, students thought an animation of rules would be helpful, but after using it to try to understand them, they no longer preferred animated graphics to static ones (Morrison, 2002). After learning, their preferences mirrored the relative effectiveness of the tools: graphics and text better than text alone, but no differences between static and animated graphics.

Events are Thought of Discretely. Given the failure of even animations that are perceptually simple, that is, animations that should be readily apprehended, let us reconsider whether animations in fact conform to the Congruence Principle. True, animations can use time to portray processes that occur in time, change over time. But how do people actually think about events that occur in time? When asked to describe ordinary events that take place in time, such as making a bed or assembling a saxophone, people describe them as sequences of steps (Zacks, Tversky, and Iyer, 2001). People think about such events hierarchically, and organized around objects and actions on objects. At the higher level of organization, each step is distinguished by a separate object—top sheet, bottom sheet, pillow cases, for making the bed—or separate object part, for assembling the saxophone. At the fine level, each step is distinguished by an articulated action on the same object: spreading the sheet, tucking in each corner, smoothing the sheet. Similarly, people think of the action of complex systems, such as a pulley, as a directed sequence of steps from beginning to end: the first pulley goes clockwise, the second, counterclockwise, and so on (Hegarty, 1992). The exact placement of the pulleys and the distances between them are ignored. Navigating the world is another everyday event that is thought of discretely, as a sequence of turns at nodes, typically landmarks or street corners (Denis, 1997; Tversky and Lee, 1998, 1999). The exact distances between nodes and the exact angles of turns are not important. The important conclusion from this is that if animated events are thought of as a sequence of discrete steps, then the congruent way to present them is as such.

This is not to say that all events are readily conceived of as a set of discrete steps, but just that many are. For some events, like patterns of weather or movements of tectonic plates, determining the boundaries of steps is not at all straightforward. Nonetheless, apprehending weather patterns from animations is difficult for novices, and even for experts (Lowe, 1996, 1999). In their actual work of predicting, experts do not use animations, preferring other forms of information (Trafton, personal communication, 2003). Moreover, descriptions of weather phenomena by experts are discretized, often by landmarks, spatial and temporal. Weather is reported for discrete time periods at discrete places.

Misconceptions from Animations. Visualizations, including animations, aren’t always a benefit. Just as for language, there are gradations of quality. Like other visualizations and like metaphors, animations can mislead; they can create misunderstandings. The experiments in which observers recognized animated motion paths correctly but reproduced them incorrectly are suggestive (Kaiser, et. al, 1992). Science educators worry that their students take visualizations too literally; for example, interpreting colors and shapes in visualizations as the colors and shapes of things that are being represented, rather than as idealization in the case of shapes or symbolic in the case of colors. There is an additional pitfall in animations, especially abstract ones, such as movements of molecules and particles. People are known to interpret movements of geometric figures as having causality, agency, and even intention. Properly staged, triangles and squares moving in a sparse environment can be seen as chasing, bullying, hiding, even talking (e. g., Gelman, Durgin and Kaufman, 1995; Heider and Simmel, 1944; Martin and Tversky, submitted). This bias to impose causality, agency, and intention to motion of abstract figures can yield misinterpretations. For example, students watching movements of various kinds of molecules, crowded balls of different colors, tumbling, coming apart, coming together, tend to see some of the molecules “pushing” others so that they will join (Tasker, 2003).

Some Hope? Even those who reluctantly accept the conclusion that animations have yet to prove themselves as superior to equivalent static graphics as teaching tools focus on the “yet,” and we are among them. Encouragement comes from the studies showing effects of animation on mental representations if not on learning. The proper animations have yet to be devised; when they are, they will prove themselves. The hope is that it’s not that animations are doomed to fail, but rather, that the proper animations have yet to be designed. Analysis of techniques used by widespread static graphics and techniques used by human explainers give clues for techniques that may improve design of animations.

Analysis of other graphics and demonstrations will also provide alternatives to animations. It will reveal the ways that static graphics convey change over time and other conceptual information that animated graphics are meant to convey. Maps are the prototype for graphics, probably the oldest and the most common. Let us first see what can be learned from them.

How Maps Communicate. Maps provide many different useful messages (e. g., Tversky, 2000). Maps of various forms have been invented and reinvented across time and space. As such, they have undergone generations of informal user-testing: people have produced them, used them, refined them to improve communication. Maps are made for various purposes, for walking, for locating significant landmarks, for driving, for weather, for hiking, for understanding spread of pollen, population, disease, for planning battles, parades, evacuations, for displaying earthquake faults, forestation, temperature, rainfall, watersheds, and more. For each of these purposes, different information is needed. A good map presents the information needed and omits the irrelevant information, which only clutters, distracts, confuses. Even after clutter is removed, some significant information may not be visible, so it is enlarged. For example, many maps meant for drivers are at a scale where roads would not be visible. So scale is violated in order to show roads or other features of importance. Maps, then, not only omit some information, they also distort other information. But maps do more than that, they also add information that is not there: arrows for battle maps, isobars for weather maps, colors for topographic maps, symbols for churches, museums, markets, railroad stations in tourist maps.

Altogether, maps, in particular, and diagrams in general, are good for showing structure. They use spatial elements to convey spatial elements, spatial relations to convey spatial relations (e. g., Tversky, 1995). But the success of various kinds of maps indicates that they do not need to portray space proportionately; on the contrary, violating metric accuracy is part of their success. Another part of their success is communicating more than just structure; maps communicate features that are not naturally visible, like precipitation and population; they express function and process, like weather patterns and earth movements. They do this by adding extra-pictorial devices like symbols, but also like arrows, boxes, brackets, and lines, devices whose geometric and Gestalt properties give clues to their interpretations in context (Tversky, Zacks, Lee, and Heiser, 2000). Importantly, among the extra-pictorial devices that maps rely on are words. Maps and other effective graphics are mixed-media. It’s hard, and ineffective as well, to do everything in pictures, even when you add visual symbols.

Highlighting, exaggerating, and distorting the relevant; omitting the irrelevant. In short, maps, and other effective static graphics, play tricks with space; they are far from 1-to-1 mappings of the world onto paper. Why not play tricks with space in animations? And why not play similar tricks with time? Most animations are linear in time; they may be in real time or they may slow or speed time, but they don’t play with temporal scale the way that common static graphics play with spatial scale. Maps don’t just expand and compress space; they also discretize it. Route maps could be analog; they could preserve distance, direction, and shape, but they do not, they are not analog, and they do not need to be analog to be useful, as they are used in contexts that disambiguate the underspecified or distorted information (Tversky, 2003). Maps play with perspective as well; tourist maps, for instance, superimpose frontal views of landmarks on overviews of roads. Animations could do that; stop and start, show views and actions selectively; change spatial scale and perspective, change temporal scale and perspective. They could combine non-linear and discontinuous uses of time with non-linear and discontinuous uses of space.

Adding extra-pictorial information. Maps and other effective graphics also add extra-pictorial features that communicate concepts that structure cannot, arrows, lines, brackets, boxes, blobs. Arrows are particularly useful: they call attention, reference, provide temporal order, indicate causality, express motion paths and motion manner, convey outcomes (e. g., Heiser and Tversky, 2002; Tversky, in press). Animations could do all that, and more.

Enriching animations in the hope of making them more effective is more than a matter of borrowing successful devices from static graphics; it is also a matter of developing devices that work in animations. One source of inspiration comes from the creativity poured into comics, cartoons, animated films, advertisements, and computer graphics. But here we look for inspiration for devices to improve animations from three research projects we are involved in.

Suggestions from Research

Diagram Narratives. In principle, animations tell stories; they are narratives. Another way to tell a story diagrammatically is a sequence of static graphics. This, too, is an ancient device; think, for example, of pictorial histories such as those in Egyptian tombs, on Trajan’s Column, in Aztec codices, on the Bayeux Tapestry, on stained glass windows. Another is a single complex graphic that has a linear reading. Contemporary examples range from scientific visualizations and assembly instructions to graphic novels (McCloud, 1994). Useful devices for enriching animations may come from studying narratives conveyed by static graphics. What kinds of stories do they tell? What devices do they rely on? In order to find answers, we conducted a survey of diagram narratives in college textbooks across a wide range of sciences, chemistry, geology, and biology (MacKenzie and Tversky, submitted). We wanted to know what sequences of diagrams are meant to convey, and how they convey it. What general themes are expressed? What steps are distinguished? How is each step portrayed? How are the links between steps established? How is abstract information conveyed? How are the depictions schematized, and what extra-pictorial techniques are used? We won’t address all those questions here. Instead, we focus on the types of narratives diagrams relate and on the extra-pictorial devices they use.

Here are most of the stories that diagram narratives tell, some appearing together:

· Change over time; cycle or process from beginning to end; implication, consequences. Some examples: the rock cycle, the circulatory system, the life cycle of a butterfly, the election process, Napoleon’s campaign on Russia, how to operate a copy machine, how to assemble a piece of furniture, how to perform a magic trick, flow diagrams, decision trees.

· Structure to function/behavior. Some examples: an engine, a cell, a leaf.

· Large to small/ whole to parts (partonomy). Some examples: a tree, a computer, a city,

· Variations of a type (taxonomy). Some examples: architectural styles, kinds of roses, diseases of the skin.

· Different views. Some examples: a building (inside or out), the ocean floor, the human body.

Enriching Diagrams with Extra-pictorial Devices. The first narrative type, change over time, is the prototypical case for animation. Large to small/whole to parts is also a natural candidate for animation, in particular, for zooming, and different views and variations of a type are naturals for panning. Graphics, animated or still, depict appearance and structure readily. Conveying change over time, function, consequences, outcomes, and other abstract content is less direct. To express abstractions, most narratives of these types are enriched with extra-pictorial symbolic graphic devices. Zooming is often conveyed by insets or a combination of brackets and lines. Brackets and lines are also used for change of perspective. These extra-pictorial devices, notably lines, arrows, boxes, blobs, and bars, often have meanings that are readily inferred from their geometric or Gestalt properties and context (Tversky, et al, 2000).

Arrows. Prominent among extra-pictorial devices are lines and arrows. Both can refer or point or label. Lines link, establish relationships. A network of lines is ideal for organizing hierarchies, such as language families or the animal kingdom or the parts of a tree. Arrows are asymmetric lines, establishing asymmetric relationships. So arrows are excellent for conveying change over time, for sequencing, for conveying causality, direction and manner of motion, consequences, outcomes, and more, perhaps too good.

The power of arrows in diagrams is demonstrated in a study in which students were asked to describe what is conveyed in diagrams of a bicycle pump, a pulley system, or a car brake with or without arrows, such as those in Figures 1 and 2. When diagrams had no arrows, participants gave structural descriptions, the spatial arrangement of the parts. When diagrams had arrows, participants gave functional descriptions, the processes and causal consequences from start to finish. Conversely, when participants were given structural descriptions, they produced diagrams without arrows, but when given functional descriptions, they produced diagrams with arrows (Heiser and Tversky, 2002). In this context, mechanical devices, people readily understand and produce arrows to convey change over time, sequence, causality.

Figure 1. Car brake with arrows* Figure 2. Bicycle pump with arrows Adapted from Mayer and Gallini (1999).

Interpreting arrows and other devices is not always immediate. Lines and arrows, and other graphic devices, have multiple meanings, as do the words whose meanings are similar, link or association or relationship. Carefully crafted context can disambiguate meanings of depictive symbols just as they can disambiguate meanings of words. Our survey, however, has turned up many examples that are not well crafted. In the diagram depicting movement of DNA in Figure 3, arrows are used to label the cell wall opening and the foreign DNA, to indicate movement of the foreign DNA, and to indicate the consequences of the process.

.

Figure 3. Movement of DNA *

* From ????

Similarly, the meanings of the arrows in the diagram of the rock cycle in Figure 4 are not at all clear. Some of the clearer ones seem to be processes that link start states with end states, for example, the arrow at bottom center seems to indicate that metamorphic rock turns into magma by melting, which in turn is a consequence of heat and pressure. What is especially confusing in this diagram is that it depicts a coherent place, albeit one with cross-sections carved out, and it is hard to ignore the more typical use of arrows in places as indicating movement.

Figure 4. Rock Cycle *

* From ????

Problematic as it is, the processes the arrows represent are labeled. Not so in Figure 5’s visualization of the nitrogen cycle. Here the boxes appear to be processors and the arrows the processes that carry the products from processor to processor. However, at least one of the processors is depicted but not named, animals, and the processes, which differ for each arrow, are rarely named.

The point of this exercise is not to show that static diagrams have their share of problems because in fact, all of these problems are fixable. The point of the exercise is to show some of the devices used to convey change over time that are not animations

Figure 5. Nitrogen Cycle *

* From ?????

Let’s look at another example of a change-over-time narrative: how to put something together. To the dismay of many do-it-yourselfers, many things that need assembly come with an exploded diagram, as in Figure 4. Exploded diagrams have the advantage of showing the decomposition while showing the whole and preserving the spatial relations of the parts. However exploded diagrams typically fail to provide the sequence of assembly, do not show the method of attachment, and all too often are at a uniform scale, so that small parts are not discernible. This diagram uses guidelines not to indicate points of attachment, which does need indicating, but rather to label parts with numbers; the key to the numbers is inconveniently elsewhere.

Figure 6. Exploded Diagram of a Motorcycle Engine.*

* From ????

Step-by-step, perspective, action. We thought users would have better ideas for visualizations of assembly (Heiser, Phan, Agrawala, Tversky, and Hanrahan, 2004; Tversky, Agrawala, Heiser, Lee, Hanrahan, Phan, Stolte, and Daniele, in press). We asked students to assemble a TV cart using the photograph on the carton as a guide. After assembling, students produced instructions for assembly. Spatial ability, as measured by a mental rotation task (Vandenburg and Kuse, 1978), accounted for not only speed and accuracy of assembly, but more significantly, for quality of diagrams. Those lower in spatial ability often only produced menus of parts, and did not show assembly. Some showed assembly steps, but with insufficient visual information about attachment operations, as in the left diagram of Figure 7. Those high in spatial ability tended to produce step-by-step perspective drawings that showed assembly actions, typically using arrows and guidelines, as in the right diagram of Figure 7. These are the key design features: step-by-step, action, perspective of action. The diagram by the high spatial participant also imposes a narrative structure on the diagram: it clearly indicates the separate steps, and it highlights the finished product. Many others included part menus at the beginning, visually indicating the starting state.

The next set of studies was aimed at evaluating the different kinds of instructions produced by the first group. New participants assembled the TV cart and then evaluated a range of instructions. Those produced by high ability participants, step-by-step action perspective diagrams, received the highest ratings by participants of both high and low ability. In yet another study, highly rated diagrams facilitated performance of low ability participants, but had no effect on high ability participants, who needed only the photograph on the box to successfully assemble the TV cart.

Figure 7. Assembly Instructions by Low Spatial Participant (Left) and High Spatial Participant (Right).

Extra-pictorial devices. Together, these findings provide strong support for the inclusion of extra-pictorial devices in visualizations. Extra-pictorial devices such as arrows, boxes, and guidelines are especially useful for conveying information that is not easily conveyed in depictions. Depictions are ideal for conveying structure; they show the essential parts of an object, device, or system in their proper spatial relations. But all too frequently, the information designers wish to convey in visualizations is dynamic in some way, the behavior of an organ or organism, the function of a device or system, the consequences of a set of operations or procedures. In order to do so, diagrams need extra-pictorial devices. Additional extra-pictorial devices our survey has uncovered include boxes or brackets combined with guidelines or arrows to indicate zooming, enlargement, or change of perspective. Animations have been used to achieve those ends as well. Useful as extra-pictorial devices are in static graphics, they are rarely used in animated ones.

Narrative structure. Similarly, animated graphics rarely superimpose a narrative structure. A narrative structure facilitates understanding and structures learning. It conveys to users or learners the beginning or initial state, it anticipates the end state, it differentiates the sequence and the steps required to achieve the end state, and it highlights the end state. Interestingly, verbal explanations, for example, those in textbooks, typically impose a narrative structure.

Demonstrations. If someone were to produce an animation to show how to assemble the TV cart, we suspect that they would animate the good diagrams, that is, they would show the parts coming together from the assembly perspective, using change over time to convey change over time. But would this be how human demonstrators would show others how to assemble the TV cart? Or would it be the most effective way of conveying assembly? The need to enrich animations is further supported by observing people demonstrating how to do something for others. We asked students to assemble the TV cart, and then to make a video showing others how to assemble it (Lozano and Tversky, submitted). One group of participants was told not to use language, as the video was meant for non-English speakers; this group relied primarily on gestures. Another group was free to use both language and gestures. A control group simply reassembled the TV cart. What distinguished assembly to instruct from pure assembly? For one thing, demonstrators made sure that the parts to be attached and the attachment actions were visible to the camera, even when this made assembly far more awkward. This parallels the earlier result on good static visualizations: they present the perspective of assembly action, showing both parts and connections. In addition, both groups made wide use of gestures, primarily pointing and exhibiting, but also modeling the assembly. They tended to point to small parts, usually connectors, and to exhibit larger parts, holding them up to the camera. As for all communication, redundancy is used to facilitate. Speaking demonstrators more often than not accompanied their speech with gestures. When allowed to use language, demonstrators accompanied their actions with clear verbal markers of steps and with caveats, things to be careful of during assembly. Those limited to gestures also marked steps and conveyed caveats, though less frequently as these are harder to express in gestures. In other words, demonstrators imposed a narrative structure on the assembly, a beginning, the initial state showing the parts, a middle, clearly marked steps, and an ending, the final state.

Remarkably, many of the critical aspects of assembly were more in evidence in the group restricted to gestures than in the group allowed to use speech as well. Gesture-only demonstrators made nearly all assembly actions visible to viewers whereas gesture and speech demonstrators made only about two-thirds of their assembly actions visible to viewers. Gesture-only demonstrators pointed and exhibited more than gesture and speech demonstrators. Both groups of demonstrators used gestures to create models, that is, three or more related gestures. The speaking demonstrators only modeled the structure of the TV cart, whereas those demonstrators restricted to gesture, modeled assembly action as often as they modeled cart structure. Frequently, the modeling was a preview of assembly action.

Parallels among Words, Pictures, and Gestures. Many of the gestures people add to assembly actions in order to demonstrate how to assemble parallel pictorial and extra-pictorial devices used in highly-rated visual instructions. Points are like lines, they can refer, label, direct attention. Exhibits also function to refer, but they can do more. Exhibits implicitly describe, they show what something looks like, just as depictions and verbal descriptions can do. Modeling structure is like structural diagrams, and modeling action is like action diagrams. Performing assembly actions so that they are visible to viewers parallels drawings that depict the perspective of assembly. Each of these communicative functions has a parallel in language as well, structure, action, perspective. Human communication is remarkably flexible. Despite differences in these modalities, many important meanings can be conveyed in words, gestures, or sketches.

The lessons to be applied to animations from demonstrations, then, are quite similar to those derived from productions of visual instructions: break action into clear steps. show the action of each step from the perspective of the viewer; give previews; add caveats; impose a narrative structure. We have already observed that animations rarely do more than show the perspective of action.

Lessons for Designing Effective Animations

We began with the bad news that animations meant to teach don’t do it better than equivalent static graphics. There are explanations, of course. Animated graphics are hard to perceive, too much change too quickly. Animated events are typically conceived as sequences of discrete steps. Nevertheless, the hope remains that the proper animation has yet to be designed. There is reason for that hope. Most animations are continuous in time, with animated time proportional to “real” time. Similarly, most animations are continuous in space. Most animations simply depict; they do not add non-depictive communicative devices, most significantly, explanations.

Analysis of successful static graphics quickly reveals that they do more than depict and that they do not map space continuously. Take a paradigmatic example, route maps, sketches designed to tell others how to get from A to B. They are sequences of actions at nodes, turns at landmarks. To convey that information, they distort and eliminate other information. Exact distances, exact angles of turns, exact shapes of landmarks aren’t needed. Nor are streets and landmarks not on the route. Sometimes extra-pictorial information is added. If static graphics play tricks with space, why shouldn’t animated graphics do the same? And why shouldn’t animated graphics play the same tricks with time?

Hints for ways to supplement and amplify animated graphics to increase their efficacy have come from a review of three projects that we are involved in. The first is a survey of static graphics meant to convey change over time and other narratives animated graphics are used to convey. Static graphics discretize the actions in time to steps. They link the steps with arrows and lines, and other extra-pictorial devices. The depictions of the steps and the uses of the arrows and lines are not always clear. Arrows, for example, have a range of meanings, referring, attracting attention, suggesting sequence, movement, change, causality, implication, outcome, and more. Frequently, a single visualization will use several meanings of an arrow or a line without disambiguating them. In a second project, we examined one domain in detail, visual instructions for assembling an object produced by experienced users. This more restricted domain also elicited extra-pictorial devices, notably arrows and lines. In this context, the meanings of the extra-pictorial devices were far clearer. The third project examined demonstrations of object assembly, contrasting them to assembly alone. Demonstrations of assembly differ considerably from actions of assembly. Demonstrations discretize actions into clear steps, each corresponding to a large object part. They supplement assembly actions with language and gestures that serve several purposes: to separate steps, to indicate parts to be assembled, to preview action, to demonstrate action and desired structure. Animations that are continuous in space and time rarely adopt these practices that humans do when teaching. It is time to “teach” animations, and these practices are have potential.

There is yet another task that animations can learn, one suggested both by the narratives accompanying animations in the work of Mayer and Sims (1994) and by the narratives in plain vanilla science textbooks. The narratives explain. Most animations just show. Showing isn’t explaining. If seeing the sequence of steps were sufficient for understanding, then schools would have an easier job.

What makes a good explanation? Many things. Explaining a process can be supplemented with analogies to other processes, with enlargements of subprocesses, with different views, with examples, with data showing consequences, with contrasting processes that differ in crucial ways. These aspects of good explanation can be done visually as well. It entails carefully designing each frame of a visualization, and linking the frames in a deliberate manner which is clear to learners. Much as a well-crafted graphic novel links its’ frames, using language as well as graphics (McCloud, 1994). Returning to maps, you can think of a good explanation as a good route from here to there, with distinctive and informative landmarks and clear paths from landmark to landmark. Like an effective route map, an animation doesn’t have to preserve temporal or spatial continuity. Just like good explanations and good maps, animations can include other views, other scales, other examples, other processes; they can use language and extra-pictorial devices to connect them. Animations can exaggerate and minimize and distort and highlight information. Animations can slow down, speed up, and stop for a moment to let learners absorb and study the current frame. There is no guarantee that shifting focus from showing to explaining will produce animated graphics that are effective, but it does look promising.

.

.

References

Betrancourt, M., & Tversky, B. (In press). Simple animations for organizing diagrams. International Journal of Human Computer Studies.

Denis, M. (1997). The description of routes: A cognitive approach to the production of spatial discourse. Cahiers de Psychologie Cognitive, 16, 409-458.

Gelman, R., Durgin, F., and Kaufman, L. (1995). Distinguishing between animates and inanimates: Not by motion alone. In D. Sperber, D. Premack, and A.J. Premack, Editors, Causal cognition: A multidisciplinary debate. Pp. 150-184. Oxford: Clarendon Press.

Edgerton, Samuel Y. (1991). Heritage of Giotto's geometry. Cornell University Press.

Hegarty, M. (1992). Mental animation: Inferring motion from static displays of mechanical systems. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 1084–1102.

Hegarty, M., Narayanan, N. H. & Freitas, P. (in press). Understanding machines

from multimedia and hypermedia presentations. In J. Otero, J. A. Leon &

A. Graesser, Eds. The psychology of science text comprehension. Hillsdale, NJ: Lawrence Erlbaum.

Hegarty, M., Quilici, J.Narayanan, N. H., Holmquist, S. & Moreno, R. (1999). Designing

multimedia manuals that explain how machines work: Lessons from evaluation of a theory based design. Journal of Educational Multimedia and Hypermedia, 8, 119–150.

Heider, F., & Simmel, M. (1944). An experimental study of apparent behavior. American Journal of Psychology, 57, 243-259.

Heiser, J., Phan, D., Agrawala, M., Tversky, B., and Hanrahan, P. (2004). Identification and validation of cognitive design principles for automated generation of assembly instructions. Proceedings of Advanced Visual Interfaces ’04.

Heiser, J. and Tversky, B. (2002a). Diagrams and descriptions in acquiring complex systems. Proceedings of the Cognitive Science Society. Mahwah,N.J.: Erlbaum

Heiser, J. and Tversky, B. (2004). Characterizing diagrams produced by individuals and dyads. In T. Barkowsky (Editor).

Kaiser, M. K., Proffitt, D. R.,Whelan, S. M. & Hecht, H. (1992). Influence of animation on dynamical judgments. Journal of Experimental Psychology: Human Perception and

Performance, 18, 669–690.

Lozano, S. C. and Tversky, B. (submitted). Communicative gestures facilitate problem solving for both communicators and listeners.

Lowe, R. K. (1996). Background knowledge and the construction of a situational representation from a diagram. European Journal of Psychlogy of Education, 11, 377-397.

Lowe, R. (1999). Extracting information from an animation during complex visual processing. European Journal of the Psychology of Education, 14, 225-244.

MacKenzie, R. and Tversky, B. (submitted). Diagrammatic narratives: Telling scientific stories effectively with diagrams.

Massironi, M. (2002). The psychology of graphic images: Seeing, drawing, communication. Mahwah, NJ: Erlbaum.

Martin, B. and Tversky, B. (submitted). Making sense of abstract events: Building event schemas.

McCloud, S. (1994). Understanding comics. NY: Harper Collins.

Mayer, R.E. & Anderson, R.B. (1991). Animations need narrations: An experimental test of a dual-coding hypothesis. Journal of Educational Psychology, 83, 484-490.

Mayer, R. E. & Gallini, J. K. (1990). When is an illustration worth ten thousand words? Journal of Educational Psychology, 82, 715-726.

Mayer, R. E. & Sims, V. K. (1994). For whom is a picture worth a thousand words? Extensions of a dual-coding theory of multimedia learning. Journal of Educational Psychology, 86, 389-401.

Mijksenaar, P. & Westendorp, P. (1999). Open here: the art of instructional design. London: Thames and Hudson.

Morrison, J.B. (2002). How goals affect evaluations of animation effectiveness. In W. Gray and C. Shunn (Editors), Proceedings of the 24th Annual Meeting of the Cognitive Science Society (p. 1022). Mahwah, NJ: Erlbaum Associates.

Muybridge, E. (1955). The human figure in motion. NY: Dover

Muybridge, E. (1957). Animals in motion. NY: Dover.

Pani, J. R., Jeffres, J. A., Shippey, G. T. & Schwartz, K. T. (1996). Imagining projective

transformations: Aligned orientations in spatial organization. Cognitive Psychology, 31,

125–167.

Pedone, R., Hummel, J. E., & Holyoak, K. J. (2001). The use of diagrams in analogical problem solving. Memory and Cognition, 29, 214-221.

Shiffrar, M. M. & Shepard, R. N. (1991). Comparison of cube rotations around axes inclined relative to the environment or to the cube. Journal of Experimental Psychology: Human Perception and Performance, 17, 44–54.

Tasker, R. (2003). Personal communication.

Trafton, G. (2003). Personal communication.

Tversky, B. (1995). Cognitive origins of graphic conventions. In F. T. Marchese (Ed.). Understanding images. (pp. 29-53). New York: Springer-Verlag.

Tversky, B. (2001). Spatial schemas in depictions. In M. Gattis (Ed.), Spatial schemas and abstract thought. Pp. 79-111. Cambridge: MIT Press.

Tversky, B. (2000) Some ways that maps and graphs communicate. In Freksa, C.., Brauer, W., Habel, C and Wender, K. F.. (Eds.), Spatial cognitiion II: Integrating abstract theories, empirical studies, formal methods, and practical applications. Pp. 72-79. N. Y.: Springer.

Tversky, B. (2003). Navigating by mind and by body. In C. Freksa, W. Brauer, C. Habel, K. F. Wender (Editors), Spatial Cognition III: Routes and Navigation, Human Memory and Learning, Spatial Representation and Spatial Reasoning. Pp. 1-10. Berlin: Springer Verlag.

Tversky, B. (In press). Functional significance of visuospatial representations. In P. Shah & A. Miyake (Eds.), Handbook of higher-level visuospatial thinking. Cambridge:

Cambridge University Press.

Tversky, B., Agrawala, M, Heiser, J., Lee, P., Hanrahan, P., Phan, D., Stolte, C. and Daniele, M.-P. (In press). Cognitive design principles for automated generation of visualizations. In G. Allen (Editor), Applied spatial cognition. Hillsdale, NJ: Erlbaum.

Tversky, B., & Lee, P. U. (1998). How space structures language. In C. Freksa, C. Habel, & K. F. Wender (Eds.), Spatial Cognition: An interdisciplinary approach to representation and processing of spatial knowledge. (pp. 157-175). Berlin: Springer-Verlag.

Tversky, B., & Lee, P. U. (1999). Pictorial and verbal tools for conveying routes. In Freksa, C., & Mark, D. M. (Eds.). Spatial information theory: cognitive and computational foundations of geographic information science. (pp. 51 64). Berlin: Springer.

Tversky, B, Zacks, J., Lee, P. U., & Heiser, J. (2000). Lines, blobs, crosses, and arrows: Diagrammatic communication with schematic figures. In M. Anderson, P. Cheng, and V. Haarslev (Editors). Theory and application of diagrams. Pp. 221-230. Berlin: Springer.

Vandenberg, S. G. and Kuse, A. R. (1978). Mental rotations. A group test of three-dimensional spatial visualization. Perceptual motor skills, 47, 599-604.

Zacks, J., Tversky, B. & Iyer, G. (2001). Perceiving, remembering, and communicating

structure in events. Journal of Experimental Psychology: General, 130, 29–58.