Table of Contents
This chapter reviews research on using animated visuals in instruction; it complements and continues the previous chapter. Special considerations in the interpretation of animated visual research are examined. Unlike research on static visuals, a very limited number of animation studies have been conducted so far. Research on two main instructional applications of animation are reviewed: presentation strategies and visual feedback from simulations. In addition, an overview of the goals and philosophy behind an ongoing research agenda is also discussed.
After reading this chapter, you should be able to:
After reading this chapter, you should be able to:
Animation is a popular and favorite effect among producers of computer-based instruction. Reviews of available courseware for education and training demonstrate its popularity (e.g., Sales, Tsai, & MacLeod, 1991). Animation is most commonly used for cosmetic purposes, with the intent of impressing rather than teaching. All too often, animation is added to CBI without serious concern for its true instructional purpose or impact. While it is easy to be critical of designers and developers, they share responsibility with consumers who tend to evaluate instructional effectiveness based on the number of frills that a package contains. This can create a cycle where the market, rather than learning needs, drives instructional design. Equally accountable are researchers who have yet to provide adequate guidance to either group. It is hoped that an attitude of shared responsibility will begin to prevail in more fully exploiting the potential of animated visuals in learning environments.
There is nothing inherently wrong with using animation for cosmetic applications, as long as it does not interfere with other lesson functions. It is tempting to say that animation scattered throughout a lesson should add to the lesson's motivational appeal. While there is a certain amount of face validity to this position, no hard evidence supports it. If animated visuals do, in fact, increase a lesson's extrinsic motivational appeal, such appeal would be based on novelty and be only temporarily effective.
Given the proliferation and power of animation production packages on microcomputers, it is expected that animation will be used even more rampantly in the future. GUI systems are increasing the ability of nonprogrammers to author dazzling graphic displays (see chapter 3). From a production perspective, animation can be dramatic and spectacular. But does this imply that learning effects should be equally impressive? Many designers and educators probably think so. Unfortunately, the research does not support such a position. It is easy to be "seduced" into believing that special effects, such as sound and graphics, must be very helpful to learners. Research, however, indicates much more complex conclusions.
The purpose of this chapter is to review the relatively small pool of instructional animation research (compared to that available on static visuals). This chapter summarizes and extends an earlier review of animation (see Rieber, 1990a). Similar to research on static visuals, early results in instructional animation research were generally negative, but also prone to confounding on many counts. Recent work has begun mapping out some of the conditions under which animation can effectively aid learning. Two main areas of research have been conducted on animation in instruction: (1) as a way of presenting information; and (2) as visual feedback in practice strategies, such as simulations. Given the available research, it seems clear that animation exerts a relatively subtle influence on learning and that many factors can further undermine this effectiveness. Despite recent advances in applying learning and instructional theory to CBI design (see Jonassen, 1988a), we know surprisingly little about some of the computer's most fundamental presentation and interactive components.
Although there are many additional applications of animation beyond presentation and practice, virtually no research is available on them. For example, it could be argued that animation provides a good way to gain the attention of a student and also to cue a student to attend to the most critical features of a screen display. As already discussed, attention-gaining is an important initial event of instruction (R. Gagné, 1985). Certainly, animation affords many practical methods of gaining and cueing attention, such as special effects during transitions between screens, (see Footnote 1) moving icons or characters (including cartoons and text), and animated prompts (such as arrows that direct attention to keywords, paragraphs, graphics, or other screen items or features). Animated objects contrast with a static background, thus increasing their prominence. (See Footnote 2)
The most direct application of animation in instruction is using it to present lesson content. Animation, with or without accompanying text, offers many opportunities for presenting or elaborating facts, concepts, and principles. As discussed in chapter 4, the processing partnership between visual and verbal information is well-established theoretically (e.g., Paivio's dual-coding theory). One could describe these instructional uses of animation as "learning-by-viewing" approaches (Reed, 1985, p. 297-298). Much of the research discussed in this chapter studied animation in this way.
Although not as cleanly definable as when used in a presentation strategy, animation is also frequently used in a wide array of interactive activities. The goal of these activities is usually to practice a recently learned skill or to acquire a new skill. These can range from highly-structured to discovery-based activities and approaches. In questioning strategies, animation is often used as visual reinforcement to student answers (i.e., a "pretty picture" as a reward for getting the right answer). As warned in chapter 4, this type of feedback should reinforce only correct answers. Any "pretty picture" given for a wrong answer should never be more reinforcing than that given for a right answer. It is surprising how often this simple rule is broken.
On the other end of the continuum are highly interactive visual displays, such as simulations, in which animation is presented as a continual stream of informational visual feedback reflecting moment-to-moment changes based on student input. Animated displays used in this way are sometimes called interactive dynamics (Brown, 1983). Simulations can be used in structured learning experiences or for open-ended, discovery-based learning. The distinction between these approaches is discussed later in this chapter, as well as in chapter 8. These examples of animation in instruction can be termed "learning-by-doing" approaches (Brown, 1983).
Some examples of successful animation used in practice activities include flight simulators, physical science simulators (diSessa, 1982; White, 1984), and programs in which students learn about musical concepts (Lamb, 1982). An entirely different set of examples comes from programming languages based on graphics, such as LOGO, in which students drive an animated turtle around the screen to draw pictures (see chapter 3 and 8) (Papert, 1980; Abelson & diSessa, 1981). As discussed further in chapter 8, real-time visual feedback during a simulation demonstrates a use of animation not easily replicated with media other than computers. On the other hand, animated presentations can be generated and displayed on various media other than computer, such as film and video.
SOME IMPORTANT CONSIDERATIONS IN THE INTERPRETATION OF ANIMATION RESEARCH
Chapter 5 discussed many issues to be considered when interpreting the results (or lack thereof) of educational research in general, and static visual research in particular. Before any graphic offers the potential for increased learning, a need for external aids to visualization must be established. For example, before evaluating the effectiveness of a picture, reviewers (Dwyer, 1978; Levin, Anglin, & Carney, 1987) have stressed the importance of first determining whether a textual passage alone elicits adequate internal imaging by students. If students adequately image internally, then, obviously, the inclusion of external visuals will probably not result in any additional learning gains. Although one could argue that adding such visuals may do no harm, which may be true, there is always the potential that unnecessary visuals may distract. Even if the text does not sufficiently induce appropriate (and necessary) mental imaging, visuals must be congruent, relevant, and consistent with the information presented in the text in order to be effective (Levin & Lesgold, 1978).
Lessons learned from static visual research are believed to be relevant for animated visuals, as well. Although animated visuals can be viewed as a subset of instructional visuals, to what extent does the research on static visuals extend to animated visuals? Put another way, what distinguishes learning from static versus animated visuals? Chapter 4 described two perceptual characteristics unique to animated visuals. The most obvious is the motion attribute. By definition, animation provides the illusion of movement. However, chapter 4 described another, more subtle, distinguishing characteristic -- trajectory, or the path of travel by the animated (or moving) object (Klein, 1987). These two additional perceptual factors must be considered in designing and interpreting animation research.
Predicting a learning effect on the basis of externally provided animated visuals depends on two things. First, animated visuals, like static visuals, must pass the test of a "need for external visualization," as described above. Second, the learning of the content must depend on understanding either changes to an object over time (i.e., motion) or changes in the direction in which the object is moving (i.e., trajectory), or both. If there is no case for this second requirement, then there would be no reason why animated visuals would aid learning anymore than static visuals. In fact, a case could be made that the additional (and unnecessary) characteristics of motion and trajectory could be distracting in some way to the learner. It would also be reasonable to expect stronger learning effects when both motion and trajectory attributes are essential to understanding a certain fact, concept, or procedure, or in solving a problem.
Unfortunately, several early reports of animation research failed to meet these requirements. In fact, two studies frequently cited as proof of the ineffectiveness of animation in instruction fall into this category. The first contained serious methodological problems in the study's overall design (Moore, Nawrocki, & Simutis, 1979). Subjects in all treatments groups were required to answer review questions after each of four lesson parts. Subjects could not proceed through the lesson until they achieved at least 85% accuracy on these review questions. Obviously, this meant that by the time subjects reached the post-test, all would achieve at least the 85% performance level. Not surprisingly, there were no significant differences between treatment groups on the post-test because of this artificially induced ceiling effect (i.e., all students learned the maximum amount regardless of treatment).
There were also several serious problems in the design and execution of a second frequently cited study (King, 1975). The materials were not sufficiently difficult, which probably also resulted in ceiling effects. The test materials were also heavily weighted to measure verbal kinds of information and thus may not have been sensitive enough to parts of the lesson demanding active visualization on the part of the students. Finally, the actual graphics used were very crude.
In addition, both of these studies used an adult population and neither provided any evidence to indicate that visuals of any type were needed to learn the material. Accepting these studies as conclusive evidence for the inability of animation to promote learning is akin to basing all design considerations on picture effects solely on the studies reviewed by Samuels (1970) (see chapter 5).
Finally, animation researchers need to design their studies to control for static versus animated visuals. Otherwise, inferences cannot be made about animation directly. For example, a very early study found positive effects for animation for factual recall, but only in comparison to a verbal presentation of the materials (Rigney & Lutz [Alesandrini], 1975). Due to the lack of proper controls, there is no evidence to suggest that the animation was any more effective than just static graphics illustrating the same material.
OVERVIEW OF AN INSTRUCTIONAL ANIMATION RESEARCH AGENDA
Most of my research to date has been devoted to understanding how animation can be used to influence learning. I will share my research on animation in this chapter, but I will leave many of the design implications of this research for discussion in chapter 8. Believing that the dramatic visual effects of animation should beget equally dramatic learning effects was a trap I fell into early in my research. Fortunately, I persevered and have made progress toward mapping out conditions that both facilitate and undermine animation's effectiveness. The history of my research, albeit short, tells an important story.
My instructional animation research has had a number of simple goals:
Based on the simple philosophy that learning is a complex set of events, the first goal was to study a variety of instructional strategies. I anticipated that different types of instructional strategies used in different combinations should interact (i.e., vary depending on the combination used). For my work, I have used Robert Gagné's (1985) events of instruction as the guiding set of lesson design principles (see chapter 2). Interactions should be expected among the five groups, or "families," of lesson components (orientation, presentation, practice, testing, retention and transfer). As a beginning point, I have focused on the two groups that generally account for most of the variance in instruction -- presentation and practice strategies.
Given earlier research indicating that people differ in their ability to form and use images as they grow older (i.e., maturation effects) (Pressley, 1977), a second goal was to study animation using very different age groups. So far, I have used students in the early to middle grades (grades four to six) and young adults (college undergraduates). Researchers all too often overlook maturation effects. What works for adults may not work for children. Results of several studies that concluded that animation made no significant difference in learning may be attributable to this point.
A third goal of my research was to study higher-level learning outcomes, such as rule-learning and problem solving. Although lower-level learning is important and, at times, prerequisite to higher-level learning, I felt there was a need for visualization research to go beyond investigating issues involving only recognition and recall tasks.
A fourth goal was to study the effects of animation in a content area not only suitable for specific visualization issues, but also generally relevant and important to people of all ages. For this reason, I chose Newton's laws of motion. Recall the two distinguishing perceptual attributes of animation discussed in the previous section: motion and trajectory. Both attributes are crucial to understanding many aspects of laws of motion. The motion attribute is obvious enough for understanding laws of motion. Understanding whether an object should be moving or not, and at what speed, is a fundamental aspect of the laws of motion. However, it is usually not enough to simply understand whether or not something is moving and at what speed. Often, an object will move, say, from left to right or right to left, depending on the forces acting on it. These issues relate to the attribute of trajectory. In fact, motion and trajectory help define many physical principles. The best-known two are probably acceleration and velocity, which are defined on the basis of speed and direction of travel.
For the purposes of presentation, laws of motion also offered an almost unlimited assortment of examples to be presented and explained, as well as contexts in which the material could be applied. Even more important, the laws of motion afford a rich variety of practice environments, including, of course, traditional question-and-answer scenarios, but also games and simulations. A broad research literature is available from science education on how people learn the laws of motion, as well as where they have difficulty. Some of this research is known as misconceptions in science (see Eylon & Linn, 1988, and Perkins & Simmons, 1988, for reviews).
Research materials were developed according to standard principles of instructional design and visualization research procedures suggested by Dwyer (1978). A basic narrative script was developed and field-tested with children and adults. Areas in the script needing additional lesson support were determined based on the level of difficulty experienced by subjects. These pilot studies also initially established acceptable estimates of the content's validity and reliability. Although the materials vary somewhat depending on the specific goals of the study, in general, the materials have the following four lesson objectives in common:
Students will be able to:
All four objectives denote application learning, or rule-learning (R. Gagné, 1985). Objectives 2, 3, and 4 are specific applications of Newton's second law (i.e., force equals mass times acceleration, or f = ma). In general, these four objectives corresponded to separate parts of the lesson script. While other objectives -- such as applications of acceleration and velocity and varying the mass of an object -- were studied at times, these four objectives remain the core focus of the learning materials.
Besides testing for the standard kinds of performance on these objectives, other data were also collected to provide a broader perspective of the influence that the animated visuals may be having on a learner. For example, cognitive perspectives on learning, as discussed in chapter 4, expect other kinds of differences in learning than only performance. Information-processing models predict that the time necessary to both encode information from short-term memory (STM) to long-term memory (LTM), and subsequently retrieve information from LTM to STM, will vary depending on a variety of factors, such as the learning outcome (e.g., fact learning versus problem solving), the level at which mental processing is occurring (i.e., shallow versus deep), and the nature of the learning and testing activities (e.g., multiple-choice questions, games, essays, simulations).
Performance data alone often do not adequately reveal information about the fluency of encoding and retrieval, whereas latency data (the time needed by a student to complete an activity) offer more information, albeit indirect, about mental processing. Although using latency data in isolation may be criticized as biased or misleading (e.g., Siegler, 1989), latency data interpreted in combination with performance data should provide a much more complete understanding of learning than using either data alone. For this reason, most of my studies have included latency data in some or all of the analyses. Some studies recorded the time needed by students to view individual frames of instruction from the tutorial, whereas other studies recorded the time needed by students to answer individual questions. In addition, other data, such as student introspections and opinions, were also often collected to see if qualitative aspects of the students' experiences matched the results of the quantitative analyses.
The presentation strategies consisted of presenting the material in a traditional CBI tutorial, generally involving four parts. The first part typically reviewed fundamental background vocabulary and concepts, such as mass, weight, and force. This part also formally presented Newton's first law of motion. The second part generally introduced Newton's second law of motion and specific applications given equal forces in opposite directions in one-dimensional space. The third part extended applications of Newton's second law to include the effects of unequal forces in one-dimensional space. The fourth part generally extended all of the concepts and principles covered up to that point, but in the context of two-dimensional space. Therefore, the tutorial was generally hierarchical in nature, that is, increasing in difficulty and making information and skills in early parts prerequisite to understanding later parts. Two basic versions of the script were produced: one for adults and one for children. In general, the script provided about 40 frames of instruction with animation embedded in about 40% of these frames.
To date, two basic types of practice strategies have been produced. The first includes traditional questions, the second includes visually based simulations. In either case, suitable activities were developed to provide students with the opportunity to practice the skills taught in the respective part. (See Footnote 3) Examples of both presentation and practice activities will be illustrated in the rest of this chapter.
Learning a Valuable Lesson Early On
Although I am not partial to reliving unsuccessful experiences, I feel there is much to gain from learning from one's errors and misconceptions. One of my earliest studies in instructional animation research is a case in point. As already mentioned, I became susceptible to the position that animation should be a potent influence on learning -- so potent, that even a little should make a difference. With this assumption in mind, I designed a study in which animation was integrated into an orienting strategy presented before each of the four lesson parts (Rieber & Hannafin, 1988). In essence, the orienting strategy presented the main principle for the ideas in the lesson part in generalized form. An example of one of the four orienting frames is shown in Figure 6.1. I expected this one frame to provide students with an "anchor" for understanding the rest of the ideas in the respective lesson part. This is, incidentally, the underlying principle behind all orienting strategies (Ausubel, 1978; Hannafin & Hughes, 1986).
In a way, I expected each of the four animated orienting frames to act as a "magic vitamin pill" that would make the rest of the instruction all the more meaningful or "nutritious." Needless to say, the study resulted in no significant differences. Although I knew that orienting activities were generally weak instructional variables in general, I thought that animation would make a powerful enough "ingredient" to turn the orientations into an effective strategy. In addition, the fourth-, fifth-, and sixth-grade students found the content very difficult to understand, which may have reduced even further the effects of the animation. This experience made me realize for the first time that the effects of animation on learning, if any, are far more subtle than I had first imagined.
Figure 6.1
An example of using animation (ineffectively as part of an orienting presentation strategy. This frame was designed to introduce students to the main principle of the lesson part that followed.
REVIEW OF ANIMATION IN COMPUTER-BASED INSTRUCTION
This section reviews evidence that is beginning to demonstrate the conditions under which animated instruction may be effective. For several reasons, this review will be confined to computer applications of animation. Some of the research, such as that based on visual feedback from simulations, is largely medium-dependent, that is, there is really no other medium other than the computer in which comparable instructional activities can be designed and studied. However, this is certainly not the case with the presentation of information that "moves," because other media, such as film and video, can be used to develop and deliver such presentations. However, the majority of research that explored the "motion attribute" from film research is dated and its generalizability is quite limited.
For example, several studies demonstrated that learning increased when using full-motion film, but primarily for procedural tasks in the psychomotor domain, such as taking a machine gun apart and putting it back together (Spangenberg, 1973). However, these results were eliminated when the instructional materials using still pictures were improved, meaning that poor instructional design explained the differences in the results more than the simply whether or not "motion" was used. (See Footnote 4) Other examples of this film research is quite useful, such as that suggesting that people with low spatial aptitude benefit from instruction containing motion sequences (Blake, 1977). There is also evidence that people's recognition memory is greater for full-motion video scenes than for static video scenes, perhaps due to the innate human abilities of perceiving movement of objects that have evolved from living in a dynamic world (Goldstein, Chance, Hoisington, & Buescher, 1982). For a thorough, yet concise, overview of instructional research of motion in films and for a general review of television research, see Chu and Schramm (1979).
In comparison to the large amount of research on the effects of static visuals (even though much of this is focused on a small number of topical areas, as discussed in chapter 5), little research is available on animated visuals, and much of what is may be confounded, as discussed earlier. In my earlier review (Rieber, 1990a), I suggested three recommendations or conclusions be drawn from the research. Although more research is now available, these three recommendations are still valid and will be used to organize the rest of this chapter. Table 6.1 summarizes the studies discussed in this section.
"Recommendation 1: Animation should be incorporated only when its attributes are congruent to the learning task" (Rieber, 1990a, p. 79).
This recommendation is obviously an extension of the design principles extrapolated from the static visual research. However, this recommendation must be interpreted on the basis of the motion and trajectory attributes discussed previously. In order for animation to have an effect on learning above and beyond that associated with a static visual, not only must a need for visualization be present, but there also must be a need to conceptualize changes to an object over time (motion) and/or in a certain direction (trajectory).
Results from studies have been mixed. One study on teaching functions of the heart found no significant differences in comparing text, text plus still visuals, and text plus animated visuals (J. Caraballo, 1985). However, the study never presented evidence that additional visual support was needed by the adult subjects to learn the material. A subsequent study also found no differences in similar treatment groups in teaching how to compute the area of a polygon, even though care was taken to validate a need for external visualization through prior field tests (A. Caraballo, 1985). However, it turned out that the animation that actually was produced did not specifically teach the mathematical rules, but only indirectly showed relationships between various geometric shapes. For example, the program demonstrated, through animation, how two identical triangles could be combined to form a parallelogram. Generally, most college seniors and graduate students would know or remember these relationships with little prompting. Thus, the addition of animated presentations of these relationships probably had little effect on learning. Also, since both studies used an adult population, the subjects may have already been able to form internal images of the content, thereby reducing any benefit of the animation.
Having the benefit of this earlier research, I designed a study to investigate the presentation aspects of animation on learning about laws of motion (Rieber, 1989). I carefully field-tested the materials, as previously mentioned, to establish both a need for visualization and to specifically pinpoint locations where visuals should be added. I also used a younger population (fourth, fifth, and sixth graders). Despite the foresight, no significant differences were found.
However, two findings provided evidence of a "smoking gun" that may have prevented the animation from doing its job. First, the post-test scores of all the students clearly demonstrated that they found the material exceedingly difficult. (See Footnote 5) Second, latency data on the time taken by students to view the animated presentations indicated that something peculiar was going on. Students actually spent significantly less time viewing frames containing animation, such as the one illustrated in Figure 6.2. The computer began recording the viewing, or processing, time of students only when the prompt to press the space bar to go on was presented. Students had to wait until the computer finished presenting text and any animated sequences before the prompt was given. I was suspicious that students were using for other tasks the time taken by the computer to execute the animation sequences. In other words, although the computer was presenting a potentially useful animated sequence from which they might learn, students probably ignored the animated sequence and used the time for other things, such as reading the screen text. Therefore, by the time the prompt to press the space bar was displayed, they were ready to move on. The difficulty of the lesson combined with insufficient cueing to the animated sequence could have been more than sufficient to confound the study.
Figure 6.2
A time lapse sequence showing the use of animation to visually elaborate a lesson principle.
Rieber, L.P. (1989). The effects of computer animated elaboration strategies and practice on factual and application learning in an elementary science lesson. Journal of Educational Computing Research, Vol. 5, Issue 4, p. 431-444. (© Baywood Publishing Company)
I designed a follow-up study to improve the materials based on this feedback (Rieber, 1990b). I changed the instructional design on two counts: I greatly simplified the materials and also added a special cueing strategy, as illustrated in Figure 6.3. The cueing strategy simply made it easier for students to pay attention to the animation and reduced the temptation to use the time of the animated sequence for other tasks. The strategy called for each presentation frame to be broken down into three, four, or five parts, or "chunks." Rather than viewing one screenful of information at a time, students viewed a chunk of screen information at a time. Students pressed the space bar when ready to view the next chunk. Presumably, students would better attend to the animated sequence because they had no reason to do anything else -- they would have already had sufficient time to read everything else on the screen.
Results of this study (Rieber, 1990b), in contrast to all the other animation studies conducted to date, clearly showed that students receiving animated graphic presentations learned more than students receiving static graphics or no graphics. There was one additional qualifier, however: this result was only found when students also received some sort of practice. (Practice was an additional factor studied, as will be discussed more in the next section.) This suggested that animation was effective, but only in the context of full lesson support. These results showed, finally, a situation where animation was a modestly effective presentation strategy.
This study was replicated again, but this time using an adult sample to see if the results would generalize to an older population (Rieber, Boyce, & Assad, 1990). No differences were found among the treatment conditions on the post-test measures. However, subjects' response times on the post-test indicated that those who received the animated presentations took significantly less time to answer the questions. This suggested that the animated presentations may have encouraged mental organization of the material as it was being learned. Increased mental organization of the content should result in faster, more confident, responses. This was exactly the pattern in the latency data of the post-test -- students receiving animated presentations needed less time to reconstruct the information as they answered the test questions. The implication is that although the adult subjects were sufficiently able to internally image, allowing all groups to achieve similar performance levels, the externally provided animated displays nevertheless aided the learning process, even though the performance measure was unable to detect such differences. Open-ended comments by students after the study matched this hypothesis. Students given animated presentations commented about their value, whereas students given static graphic commented that "examples of moving balls and kicks were needed" (Rieber, Boyce, & Assad, 1990, p. 50). Students given all-text versions commented that "pictures and graphics were needed" (Rieber, Boyce, & Assad, 1990, p. 50).
A more recent study shows considerable evidence of both the range and limitations of adults learning from animated presentations (Mayer & Anderson, 1991). Students were taught how a bicycle pump works. In three separate experiments, some students watched only an animation of the principles, others heard a narration of the same information but without pictures, and others saw both the animation and heard the narration either together or with the narration coming before the animation. Students given the animation along with the narration significantly outperformed students who either in isolation watched the animation or heard the narration or who heard the narration right before seeing the animation on the problem-solving tasks. Even more important, the animation without the verbal description was completely ineffective, as students in this treatment compared equally with students provided no instruction at all. Consistent with Paivio's dual coding theory described in chapter 4, learning from animation, like any visual, is best when paired with appropriate verbal support because of the increase to both representational and referential encoding.
Figure 6.3
An illustration of a special cueing strategy
which "chunked" each frame of instruction into verbal
and visual parts.
Rieber, L.P. (1991) Effects of Visual
Grouping Strategies of Computer-Animated Presentations on Selective
Attention in Science. Educational Technology Research &
Development, Vol. 39, Issue 4, p. 5-15 (© Associations
for Educational Communication & Technology.
This series of studies has begun to shed some light on some of the conditions necessary for animated presentations to aid learning. As mentioned at the onset, the demands of the learning task must match the three attributes of animation (visualization, motion, and trajectory) in order for learning to occur. However, this is a necessary, but not sufficient condition for learning. Other factors can undermine the effectiveness of animation. Some of the factors indicated by this research include: exceedingly demanding learning tasks, poor instructional design, and the inability of students to focus on or attend to the information contained in the animated display. This final intervening factor suggests the next recommendation.
"Recommendation 2: Evidence suggests that when learners are novices in the content area, they may not know how to attend to relevant cues or details provided by animation" (Rieber, 1990a, p. 82).
Based on the previous research, there seemed to be evidence that the "chunking" strategy shown in Figure 6.3 helped students focus on the animated sequence. However, the purpose of the Rieber (1990b) study was not to investigate the effects of this particular cueing strategy. Instead, the strategy was used throughout the instructional design of all the treatments, even those containing static graphics or no graphics. For this reason, I designed a study to directly test the hypothesis that this strategy did, in fact, account for greater selective attention of the animated information on the part of the students.
The study compared two versions of two visual treatment groups (static and animated visuals) (Rieber, 1991a). One version presented one screenful of information at a time, as was used in the Rieber (1989) study. This method can probably be considered the traditional approach in CBI design. The second method used the chunking strategy from the Rieber (1990b) study (see Figure 6.3). Results showed that students in the animated grouped condition performed significantly better on the post-test than students in either of the two static visual treatments (grouped or ungrouped). Post-test scores of the students in the animated ungrouped condition were not significantly different than any of the other three conditions (Rieber, 1991a). This study provided good preliminary evidence that the animated presentations would only be more effective than static visuals when students are properly cued to the information contained in the animated sequence. In the study by Mayer and Anderson (1991), animation presented without any verbal support was completely ineffective, indicating that students were either unable to appropriately focus on or to understand the most important visual parts of the presentation.
The implications of this second recommendation are easy to overlook when designing animated visual displays. Designers and developers forget that they become content experts of the materials they produce. Information contained in an animated sequence, though wonderfully obvious to them, may be totally overlooked by students. Even if students appear to be attending to the surface-level features of an animated display, they still may be unable to draw out, or "read," the information contained in the animation. Why don't students pick up on the information in a seemingly well-produced animated display better or more frequently? The answer may lie in the fact that students are probably not accustomed or trained in interpreting animated information, perhaps because much of the animation they view is meant to appeal to their affective domain, such as video cartoons. This research indicates that students must be sufficiently cued and guided in order to take advantage of the potential learning effects of animation. I believe that the "chunking" strategy used in my research is only one of many possible cueing strategies that should prove effective.
A complex study that investigated the use of graphics to teach algebra word problems (Reed, 1985) suggested that students who are beginners in an area may have great difficulty perceiving differences from animation when only required to view the displays. The study involved a series of four iterative experiments (meaning that the results of each experiment were used to improve the design of the next). The animated displays were only effective when paired with an interactive strategy that forced students to attend to critical features of the animated display. A replication of this study (Baek & Layne, 1988) provided additional evidence that students need external cueing in order to learn from animated displays.
White (1984) has described instances when students' misconceptions of a content may interfere with their perceptions of what is actually happening in an animated display. For example, if your personal "theories" of physical science tell you that an object should be moving at a constant rate, you will probably misinterpret or ignore motion cues of an object that is actually accelerating at a slow rate. Again, in contrast, the designer or expert may see these differences as obvious. As discussed in chapter 4, perception is a function of prior knowledge or experiences (described as top-down processing in chapter 4).
"Recommendation 3: Animation's greatest contributions to CBI may lie in interactive graphic applications" (Rieber, 1990a, p. 82).
This final recommendation from my earlier review (Rieber, 1990a) represents a Pandora's box of issues and applications of animation in instruction. The recommendation speaks to interactive activities in which animation plays an important role. The most obvious examples are visually based simulations, such as flight simulators, where animation is used to represent visual feedback from the artificial world modeled by the computer. In these applications, one cannot study animation per se, but only the activity within which the animation is contained. It is therefore virtually impossible to cleanly extract the effects of animation because its effects are contextually bound to the activity. For consistency and simplicity, I have repeatedly referred to such highly interactive activities as practice strategies.
This is an area in which I have just begun to do systematic research. As a first step, I have taken the position that it is more useful to compare design philosophies than small variations within a single activity. For example, in some of my work I have compared designs based on behavioral orientations, such as questioning strategies, to those based on cognitive orientations, such as visually based simulations. Although the "behavioral versus cognitive" label may be an oversimplification, it nonetheless remains a useful distinction. As discussed in chapter 4, each philosophy makes vastly different assumptions about human learning. My goal is not to divide the positions further, but merely to resolve and better understand differences in their applications to instructional design.
Relevant and sustained student interactivity is one of the most critical features of instructional design (R. Gagné, 1985; Gagné, Briggs, & Wager, 1992; Jonassen, 1988b). Successful practice strategies, such as questioning techniques, have a long history, especially for lower-level learning such as recall (Anderson & Biddle, 1975; Hamaker, 1986). Practice enhances learning in these situations by increasing overt attention to and rehearsal of relevant lesson information, combined with positive reinforcement and informational feedback (Kulhavy, 1977; Schimmel, 1988; Wager & Wager, 1985). Traditional questioning strategies have been successfully applied to CBI (e.g., Hannafin, Phillips, & Tripp, 1986), but they tend to quickly become monotonous or boring. In addition, they rely heavily on extrinsic motivation, such as rewards and reinforcers, to be successful.
Practice strategies that promote higher levels of learning demand different design assumptions (Salisbury, 1988). Learning is promoted by presenting problems or conflicts that encourage a student to use novel and original strategies, such as hypothesis-testing, to derive solutions. Cognitive psychology suggests many factors that need to be considered in designing practice strategies. These include, but are not limited to, meaningful contexts based on a learner's prior knowledge and experiences, issues related to comprehension monitoring, and intrinsic motivation (Craik & Lockhart, 1972; Keller & Suzuki, 1988; Lepper, 1985; Malone, 1981; Ross 1984).
I and my colleagues have studied visually based simulations, structured in various ways, to teach Newton's laws of motion. In most studies, students were given varied control over an animated "starship." Students manipulated the direction and frequency of forces acting on the starship, as shown in Figure 6.4. In the experiments, the simulation was used as a strategy to practice the information and skills learned in an accompanying tutorial. These simulation activities were compared to the more traditional questioning activity, as well as to a "no practice" control.
When studied with children, those given the simulation outperformed students in the no-practice control whereas those given the questioning technique did not (Rieber, 1990b). In addition, a follow-up study provided strong preliminary evidence about the intrinsic motivating appeal of the activities (Rieber, 1991b). Students overwhelmingly chose to return to the simulation when given total freedom of choice at the end of the experiment. The questioning activity and a word find puzzle were among the students' other choices. The puzzle activity was deliberately meant to be strong competition for the simulation. Not only do children of this age group (fourth graders) traditionally find such puzzles a lot of fun, but the experiment allowed them to use the puzzle as a general "escape" from any "school-like" features they may have associated with the computer materials. Measuring intrinsic motivation is tricky business, but such free-choice methods have a credible history. Choosing to participate in an activity when all external pressure to do so has been removed is generally known as continuing motivation (Maehr, 1976; Kinzie & Sullivan, 1989).
Figure 6.4
A snapshot of the screen during an episode of a structured simulation used as a practice strategy. The "starship" is under student control. This simulation is structured in various ways. For example, the starship spins in 180 degree increments, resulting in one-dimensional motion.
Rieber, L.P. (1990) Using Computer Animated Graphics in Science Instruction with Children. Journal of Educational Psychology, Vol. 82, No. 1, p. 135-140. © 1990 by the American Psychological Association. Reprinted by permission of the publisher.
When studied with adults (Rieber, Boyce, & Assad, 1990), subjects performed equally well given either the simulation or the questions -- both groups outperformed the no-practice control. However, the simulation group took significantly less time to answer post-test questions than either the question or no-practice groups. Again, this latency data suggests that the simulation activities may have aided students' organization of the material, resulting in a decrease in retrieval time.
Research on Inductive Learning
My most recent work has been studying how people learn from the simulations with and without the use of accompanying tutorials. Again, my purpose is to compare design philosophies within CBI, which in this case is the difference between deductive and inductive learning. Deductive learning involves the most traditional approaches to education, such as presenting the rule for a concept (e.g., evergreen trees) with lots of examples and nonexamples, and providing plenty of practice. Inductive approaches involve "discovering" general rules or concepts through constant and varied interaction with specific cases (e.g., repeatedly walking through the forest until you notice different kinds of trees). The way children (and adults) learn complex "dungeon and dragon" video games are good examples of inductive learning.
A good illustration of the difference between deductive and inductive approaches is learning how to swim. A deductive approach would be the Red Cross method of carefully teaching the prerequisite skills to people step by step, such as breathing, kicking, and arm techniques, first in shallow water and then in deep. Deductive methods determine learning goals and the best way to achieve them in advance. Of course, deductive methods can also be boring and routine, and teaching strategies often begin to resemble one another.
The inductive method is similar to throwing someone into the deep end and seeing what happens. If the person learns how to swim, the skill is probably learned for life because of the meaningfulness of the experience. Of course, there is also the danger of drowning on your first attempt or surviving just to be afraid of water for the rest of your life!
In daily life, people use a combination of deductive and inductive strategies. Learning how to do simple plumbing or electrical jobs around the house are good examples. If you are confident, you may just take a device apart to see if you can find what is wrong. Other times, you may consult a "how-to" book for step-by-step instructions.
Research conducted so far with adults on learning inductively from computer- and visually based simulations has been mixed. For example, a "stepping stone" study looked at the use of visually based simulation activities as orienting and practice activities combined with tutorials on learning about acceleration and velocity (Rieber, Boyce, & Alkindi, 1991). Examples of the simulation activities are shown in Figure 6.5. In general, the activities were ineffective as an orientation for later learning experiences. Similar to previous studies, the simulation activity was generally useful as a practice activity, but not under all conditions. For example, the activity was effective as practice when subjects were tested on near transfer tasks (i.e., questions that closely matched the context in which learning occurred). However, the effect disappeared when tested on the same content using novel contexts (i.e., far transfer). However, the study was partially confounded by the complexity of the material. In post-experiment surveys, students stressed that they found the principles of acceleration and velocity very difficult to learn. The survey data also seemed to indicate that students were uncomfortable with the open-ended simulation activities -- they expected more structure.
Figure 6.5
Two examples of simulations using game-like
features that allow students to interact with the principles
of velocity and acceleration.
Rieber, L.P., Boyce, M., & Alkindi,
M. (1991). The effects of computer-based interactive visuals
as orienting and practice activities on retrieval tasks in science.
International Journal of Instructional Media, Vol. 18, No. 1,
p. 1-17.
This study, as well as many informal experiences, has given me some indication that adults are generally very uncomfortable with open-ended, discovery-based activities, at least when they perceive the learning environment to be formal or "school-like" (such as in the case of participating in a research study). Adult educators have long echoed similar messages (Seaman & Fellenz, 1989). A recent study directly compared deductive versus inductive strategies, again using adults as subjects (Rieber & Parmley, 1992). Subjects given a structured simulation activity without a tutorial performed as well on performance measures as any of the conditions that included a tutorial. However, subjects given an unstructured simulation performed no better than subjects given no instruction at all. Figure 6.6 illustrates an example of a "structured" simulation, and Figure 6.7 illustrates an "unstructured" simulation. A simulation was defined as structured when it had a clear goal and guided students through stages of the skill to be learned. Unstructured simulations were much less goal-oriented and fully immersed subjects in all of the physical principles of the lesson.
Subjects were also asked to rate their response confidence as they answered the post-test questions; that is, how confident, on a scale of 1 to 5, they were that they were answering correctly. The most confident students in the experiment were those in any of the conditions containing a tutorial. However, the next confident were the subjects given the structured simulation/no tutorial treatment. Therefore, even though this group performed as well as the tutorial groups, they did not feel as confident in their learning. The lowest in confidence were subjects in the unstructured simulation and, not surprisingly, students who were given the test without any instruction.
Figure 6.6
An example of using a structured simulation
activity to inductively learn about laws of motion. When structured
in ways like this, strudents were able to learn as much with
or without formal tutorials.
Rieber, L.P. (1991). Animation, incidental
learning, and continuing motivation. Journal of Educational
Psychology, 83(3), 318-328. Copyright 1991 by the
American Psychological Association. Reprinted by permission of
the publisher.
There is also a wealth of related research in this area, such as that dealing specifically with the design of simulations (Alessi & Trollip, 1985, 1991; Atkinson & Burton, 1991; Orbach, 1979; Reigeluth & Schwartz, 1989). In fact, I designed the simulation activities to follow closely the earlier development work by diSessa (1982) and White (1984), which is based on using simulations as "microworlds" for learning physics. The constructs of simulations and microworlds are the basis for the discussion in chapter 8.
Figure 6.7
An example of using an unstructured simulation activity to inductively learn about laws of motion. Although minimal goals were established, adult subjects were not receptive to such open-ended activities. The "verdict" on whether this pattern will be found with children is still open and being researched.
Research on Learning Incidental Information from an Animated Display
An area of research somewhat related to inductive learning is incidental learning, or learning that occurs without deliberate attempt by the instruction or teacher (Klauer, 1984; Lane, 1980). Traditional instructional design, upon which most of CBI is based, has been primarily concerned with intentional learning, or that specified by carefully chosen and predetermined instructional objectives. Proponents of incidental learning accept the premise that a wide variety of learning is continually in progress, only some of which is anticipated. Research has usually indicated tradeoffs between intentional and incidental learning; that is, increases to one kind of learning usually means decreases to the other.
In an early study (Rieber, 1990b), I had some preliminary evidence that students were extracting information from an animated presentation other than what was intentionally being taught. I decided to test the hypothesis that students might be able to learn incidentally from animation and to see if there were any consequences to intentional learning (Rieber, 1991b). Students were given a tutorial on a simple application of Newton's second law, where the acceleration of an object with constant mass varies depending on the size of the force that is applied to it. The larger the force, the larger the initial acceleration and the faster the ball ultimately goes. This application was an intentional learning outcome.
Figure 6.8
An example of using animation to present incidental information to students through the motion attribute.
Rieber, L.P. (1991). Animation, incidental learning, and continuing motivation. Journal of Educational Psychology, 83(3), 318-328. Copyright 1991 by the American Psychological Association. Reprinted by permission of the publisher.
However, students given animated presentations were also exposed to another application of Newton's second law, where the mass of an object varies but the force remains constant. Through animation, students were shown the consequences of what happens when you apply the same size force to objects of different mass, such as a concrete block and a soccer ball, as illustrated in Figure 6.8. This application was an incidental learning outcome, meaning that no formal attempt was made to actually teach the application -- it just happened to be part of the animation. In fact, the purpose of the animated sequence was to promote the intentional learning goal.
Results were quite startling. The fourth graders given the animated sequences successfully extracted this incidental information and applied it in appropriate ways. Furthermore, there seemed to be no obvious decrease in their intentional learning. However, there were consequences to this "extra" learning. Not only were the students able to apply this incidental information to appropriate contexts, they also applied it to inappropriate contexts. They incorrectly used the information to help solve problems dealing with the concept of gravity, whereas students only given static graphics were not prone to such interpretations. This "good news, bad news" story shows that students are constantly processing information in a variety of ways. Sometimes, their interpretations are constructive and relate to a set of larger goals; other times they may be building misconceptions.
Some Final Comments about Animation Research
As you can see, the available research on the effects of animation on learning is quite small. That fact has influenced me to be more systematic in my research agenda in order to efficiently investigate a wide range of issues. Other work, though largely nonexperimental, has been done on the instructional effectiveness of animated graphic displays. For example, Margaret Withrow (1978, 1979) and her associates have successfully used computer animation for languaging activities with hearing-impaired students. Other work includes motion perception research (e.g., Proffitt & Kaiser, 1986; see also the history of apparent motion research in chapter 4), testing (Hale, Okey, Shaw, & Burns, 1985), learning geography (Collins, Adams, & Pew, 1978), and understanding three-dimensional orthographic drawings (Zavotka, 1987).
Given the limited research, designers and developers should cautiously and prudently interpret and apply the research results. It is hoped that much more research will be forthcoming, especially for presentation issues, as the visualization community begins to apply its high-end computer graphics systems to instructional issues. A whole range of questions related to two versus three dimensions, texture, color, and lighting and shading, remain largely unexplored (see Tufte, 1990 for discussions on designing graphics that escape from "flat land," for example). The ending from my earlier review of animation is still very relevant here: "CBI designers are faced with a curious dilemma. They must resist incorporating special effects, like animation, when no rationale exists, yet must try to educe creative and innovative applications from the computer medium" (Rieber, 1990a, p. 84).