Table of Contents
OVERVIEW
This chapter presents a review of research dealing with the effect of pictures and other graphics on learning. It concentrates on research of static graphics, whereas the next chapter deals with animated graphics. A simple research evaluation model and instructional visual research taxonomy are first discussed to help guide readers through some of the issues related to interpreting educational research. An overview and summary of major published reviews is presented chronologically in order to provide a historical context for the research literature. In general, research suggests that pictures do not always help learning, and, at times, may actually be counterproductive. The chapter concludes with a series of general instructional design principles to help guide practitioners on the basis of available research evidence.
OBJECTIVES
Comprehension
After reading this chapter, you should be able to:
Application
After reading this chapter, you should be able to:
Given the widespread use of illustrations and other types of graphics in instruction, one would think that there would be definitive research literature to either support or dispel their usefulness. Although the use of pictures as an instructional aid has been a very popular research issue, the literature is far from definitive and, at first glance, can even appear contradictory. For example, researchers studying the effects of pictures in prose learning prior to 1970 concluded that pictures often did not aid children's learning and was even distractive at times (Braun, 1969; Samuels, 1967, 1970). Research conducted since 1970 has been more supportive of instructional visuals, not because students are somehow different now, but because we have a better sense of the conditions under which visuals work. This implies that not only do a set of conditions exist, but that pictures will not, and should not, help learning in every instance.
It is easy to be misled into thinking that relatively simple questions like "Do visuals help people learn?" or "Will color make visuals more effective?" should have equally simple answers. An analogous question would be "Is a hammer a good tool to use?" The answer is sure, sometimes, but it all depends. It's great for hammering and pounding nails, but pretty lousy for cutting hair. Researchers in recent years have done a much better job of unraveling and explaining what the effectiveness of instructional visuals depends on. The purpose of this chapter, and the one that follows, is to present a simplified overview of instructional visual research.
This chapter concentrates on the large pool of research conducted on static visual research, a pool which is largely media independent (i.e., not more or less related to one instructional medium, like the computer, than another). The next chapter focuses on the relatively small group of studies dealing with animation, particularly that associated with computers. Both chapters deal exclusively with experimental research that strives to find some causal relationships between instructional variables, such as graphics, and learning. In contrast is correlational research that only looks for relationships between variables, but cannot tell whether one variable (e.g., a picture) caused a change in the other (e.g., learning). Some studies try to bridge the two by studying the relationship between, for example, learner aptitudes and instructional variables (Cronbach, 1957; Snow, 1977). This is known as aptitude-by-treatment interactions (ATI) and has been used to some degree, in research on instructional visuals (Dwyer, 1978). A third branch of research methodology, best known as naturalistic inquiry, is virtually without representation in research on pictures in instruction. Whereas experimental and correlational research is largely quantitative in nature, methodology based on naturalistic inquiry is predominately qualitative in nature. (See Footnote 1)
There have been many recent reviews of research on static instructional visuals. Rather than just offer another review, this chapter will present a "meta-review," or review of reviews. Furthermore, this review will be presented chronologically, in the hope of providing an historical context. Before beginning, it is important that we have a way of understanding and interpreting the research. To this end, the next section will present a simple model useful in evaluating and interpreting research. This evaluation model, though useful for understanding any educational research, is particularly helpful in knowing when research is either a friend or a foe to instructional designs that incorporate graphics.
INTERPRETING RESULTS OF INSTRUCTIONAL VISUAL RESEARCH
Readers are cautioned at the onset to be suspicious of any experimental research result in the social sciences, for example, educational research. Unlike much research in the hard sciences, control of all extraneous variables in the social sciences, except that which you are actually studying (called the independent variable), is not only extremely difficult, but actually can be criticized for jeopardizing the generalizability (or usefulness to the field in general) of the results. This dilemma is summarized by the distinction between control and maintenance of internal and external validity.
For example, if one wants to study a question related to the effectiveness of illustrations, then internal validity calls for only that variable, visuals, to be allowed to vary between groups in the study: for example, one group is given visuals and another is not. A high control over the study's internal validity, or control of the conditions that actually influence the experiment, makes the researcher confident that if one group learns more than the other, it must be the visuals that are "causing" the difference. However, this type of learning is usually criticized as being too artificial or "sterile" to be generalized to real training situations in which learning occurs in the midst of many mediating factors. Generally, the higher the internal validity, the lower the external validity. Conversely, studies that try out variations of instructional variables in "the real world" under "real" conditions unfortunately risk the criticism that the lack of internal controls makes interpretation of any resulting differences difficult or impossible. The issue of balancing internal and external validity can never be totally resolved. However, this is an issue that must always be in the mind of anyone who intends to interpret research results (see Borg & Gall, 1989, for a discussion of sources of validity in relation to educational research).
There are, however, many more issues that must be taken into consideration when trying to interpret educational research. Learning involves a dynamic interplay of many variables, most of which vary even during the learning event itself. A simple model of the many interdependent instructional variables consists of four elements (Jenkins, 1978): the learner; the learning activities; the learning materials; and the testing environment. Though this model is simple, it illustrates a complex set of relationships among general groups of factors that must always be kept in mind when exploring questions about learning, understanding, and remembering. In other words, understanding research requires that each of the four variables be recognized and interpreted in light of the other three. This model likens educational research to a game of pick-up sticks. One cannot manipulate one of the variables without affecting the other three. Figure 5.1 visually represents the dynamic interplay of these variables into a model useful for the evaluation of educational research.
When reading and interpreting research results, many questions should arise based on the interplay of this model. For example, consider a study that reports significant results in favor of visuals. (See Footnote 2) If children were used as subjects, an important question for the practitioner might be whether the results generalize to an adult population. Similarly, if a study reported significant results related to a test of factual recall, do the results generalize to problem-solving situations? If the study reported results with the use of simple line drawings, do the results generalize to color photographs? If the study used an individualized learning approach where learners had control over the pacing of the lesson frames, do the results generalize to group instruction where the visuals are externally paced (such as a video)? It is easy and tempting to believe that a study's results will completely generalize to your situation, especially when the results seem to support your position. Obviously, no study can ever completely match an applied circumstance, but it is vital that the main issues and procedures of a study dovetail with one's purposes and needs sufficiently before research results can be used to support or refute one's design decisions.
Figure 5.1
Some criteria useful in the interpretation of educational research. The four points of this model are interdependent; that is, you cannot change one without affecting the others. Therefore, each point must be intepreted in light of the other three.
A useful analogy for understanding and interpreting research on instructional visuals is the construction of a brick wall. Any one research study represents but one brick. A collection of unrelated studies does not begin to answer a question like "Do visuals aid learning?" any more than a pile of bricks defines a wall. Similarly, the results of any one study must be viewed in relation to the rest of the literature, just as any one brick's contribution to the wall can only be understood when looking at the entire structure. Some studies, like some bricks, become the foundation for others. Other studies, such as replications of previous work, act as functional facades, supporting and confirming what we already know. Obviously, no one study can answer all questions, even if all of the questions are known. At its best, educational research is an organized collection of studies, each of which connects to the ones that precede and follow it.
Guided by the evaluation model in Figure 5.1, we can construct our own "brick wall" of research on instructional visuals in the form of a matrix or taxonomy. For our purposes, the types of graphics (i.e., representational, analogical, arbitrary) relate directly to learning materials. The instructional functions (i.e., cosmetic, motivation, attention-gaining, presentation, practice) and the domain of learning in which each is applied (i.e., verbal information, intellectual skills, cognitive strategies, affective, psychomotor) relate to the learning activities. For example, the crossing of graphic types and learning outcomes results in a matrix with 15 cells or "bricks," as illustrated in Figure 5.2. By adding the five instructional functions discussed in chapter 2 (three cognitive and two affective) and the most obvious psychomotor function (i.e., demonstration of procedural, or "how to," tasks), the matrix becomes three-dimensional, as shown in Figure 5.3. Legitimate research can and should be conducted in any one of the 36 cells. The matrix is further complicated by considering the remaining two issues suggested by the evaluation model within each cell (or brick) -- the nature of the learner and testing components, as well as other issues, such as Gagné's events of instruction. This matrix, in essence, creates a rather thorough taxonomy for general goals related to research of instructional visuals. This research taxonomy helps to clarify where research has been done and to what instructional designs the research applies. The taxonomy also guides work remaining to be done.
Figure 5.2
Simply crossing the various domains of learning with the illustration types describes some of the areas in which research has been conducted.
Figure 5.3
Adding the consideration of instructional functions to the matrix in Figure 5.2 provides a fairly thorough research taxonomy to help interpret existing research and to show where more research needs to be done.
OVERVIEW OF STATIC VISUAL RESEARCH
The question is, then, what do we know about visuals within areas suggested by the taxonomy of Figure 5.3? Simply put, we know a lot about a little. The lion's share of the research literature deals with the use of static visuals to supplement the presentation of textual information in reading, commonly known as prose learning. The remainder of this chapter summarizes this pool of research. The sequence of instructional visual research and its subsequent interpretation unfolds in an interesting and not so predictable way. For this reason, this overview will chronologically present a series of the most serious and notable reviews available. Considered together, these reviews demonstrate the frustration of early and mostly negative results of using visuals to teach, followed by more positive results. The latest reviews point to a greater understanding of how instructional visuals may contribute to learning when part of global instructional strategies. Again, this chapter focuses only on static visual research. The role of animated visuals is discussed in the next chapter, especially in relationship to computer-based instruction.
Distraction Effects of Pictures: Review by S. Jay Samuels, 1970
One of the earliest serious reviews of instructional visual research was published by Samuels (1970) and began with the following sentence: "If fish were to become scientists, the last thing they might discover would be water" (p. 397). Samuels chastised researchers for failing to investigate "the ubiquitous use of illustrations in books for beginning reading instruction" (p. 397). This admonishment was directed at the failure of researchers to study one of the most widely adopted and touted instructional aids in use -- pictures. Samuels reviewed studies on the basis of picture effects on three learning outcomes: learning to read, comprehension, and attitudes.
The first category of experiments -- learning to read -- studied the use of pictures in teaching simple vocabulary to children. Most of these studies were strictly behavioral in nature and used a pair-association task for the learning activity. This usually involved presenting a series of printed words with or without accompanying pictures of the words. In almost all of the studies Samuels reviewed, there was usually either no difference between the picture and no-picture groups, or students performed better with no accompanying pictures. It was believed that children's attention was naturally drawn to the stimulus that most easily resulted in the correct answer during the learning trial (i.e., the picture), as suggested by the principle of least effort (See Footnote 3) (Underwood, 1963). Therefore, students were usually unable to recall the word when tested only with the written word. Samuels concluded that pictures generally interfere with young children's learning of sight vocabulary. Usually, children classified as poor readers were most susceptible to this interference effect.
The last two categories of studies reviewed by Samuels, comprehension and attitudes, unfortunately posed many problems, making interpretation ambiguous at best. First, too few studies were represented in each case -- two in the case of attitudes. Second, the studies that were represented appeared prone to confounding on many counts. For example, many of the comprehension studies tested for memory, not comprehension. The quality of the design of the studies is also easily questioned. The Samuels' review, therefore, largely left unanswered the questions of whether pictures affect reading comprehension and attitudes.
The value of the Samuels' review relates to its evidence of the potential distracting nature of visuals. Samuels concluded that students, usually those with below-average reading skills, had difficulty shifting their attention from a picture to a written word because the picture required less effort. Anyone who has ever tried to assemble a children's bicycle late on Christmas Eve knows of the temptations to skip the written directions and defer to the accompanying pictures. However, it is possible to overlook relevant information in the picture or misunderstand relevant information.
A classic demonstration of the distractive potential of pictures is a pair of studies conducted by Willows (1978). Second- and third-grade children were presented with words superimposed on either related or unrelated pictures. A no-picture group acted as a control. Results showed that the children read more slowly and with less accuracy in the presence of pictures than without pictures. Furthermore, unrelated pictures produced more interference than related pictures. Willows concluded that "the children either consciously or automatically and unconsciously attempted to use the pictures as clues to the meanings of the words printed on or near them" (p. 261). Similar to Samuels, Willows found that younger, less-skilled readers were more susceptible to this interference effect and also attributed the result to the principle of least effort. Willows further suggested that the results from his word-picture association experiments generalized to reading prose or narrative, and discussed the distraction effects in relation to designing books for children:
In most books used for reading instruction, the illustrations which accompany a story are complex and include representations of many components of the text. If a child comes to a word he already knows, then the pictures in the periphery are superfluous and probably distracting. If he does not know a word and looks to the picture for a clue to its meaning, he may well be misled by those aspects of the picture which are not closely related to the meaning of the particular word he is trying to decode. (p. 261)
Describing the Conditions Under Which Pictures Facilitate Learning
Research conducted since the Samuels' review points to the positive effect of pictures in prose learning. Recent findings suggest that pictures can exert strong positive influences on learning, given certain conditions. Further evidence suggests that children's dependence on pictures decreases with age. As children grow older, they become better able to produce their own internal images (Pressley, 1977).
For example, Guttmann, Levin, and Pressley (1977) read stories to kindergarten, second-grade, and third-grade children and presented either sets of pictures that fully illustrated the content of the stories (imposed pictorial conditions), sets of pictures that illustrated the story but omitted all information contained in the questions (partial-picture conditions), or no pictures. Subjects in the partial-picture condition were explicitly instructed to form mental images. Children in the control condition neither saw pictures nor were instructed to internally construct them. Results showed that even kindergarten children learned more when presented pictures illustrating the story; however, the partial-picture group showed no difference from the control group. Second-grade children were more able to use the incomplete pictures and seemed to be able to construct an internal image if given a partial picture, but incomplete pictures still did not produce significant differences in learning when compared to that of the control group. Third graders in all three experimental conditions, on the other hand, significantly outperformed control subjects. This study suggests the developmental importance of imagery ability. Imagery abilities and skills, like other cognitive processes, probably develop over time. Other studies support this contention (Shimron, 1975; Lesgold, Levin, Shimron, and Guttmann, 1975).
Other research, although supporting the claim that children depend less on outside images as they grow older, demonstrated that pictures can decrease the difficulty of prose material for older children. Levin and Divine-Hawkins (1974) demonstrated that fourth-grade children do not automatically construct images, although they are capable of doing so. This finding has led to many examples of successful training of subjects to form mental images (see, for example, Lesgold, McCormick, & Golinkoff, 1975; Pressley, 1976).
Review by Joel Levin and Alan Lesgold, 1978
Since the late 1970s, many reviews of instructional visual research have been published that point to positive effects of pictures on learning. One of the earliest was a review by Levin and Lesgold (1978), which described "abundant empirical evidence to document the positive value of pictures" (p. 233). They reported that consistent learning gains were to be expected by the addition of pictures when the following five ground rules were followed: (1) the prose passages were presented orally; (2) the subjects of the experiments were children; (3) the passages were fictional narratives; (4) the pictures overlapped the story content; and (5) the learning was tested by factual recall.
The purpose of employing these ground rules was to try to sift through the hundreds of studies available for some consistent set of conditions that could be examined, interpreted, and debated. These ground rules also made it possible to exclude studies that, although published, were poorly designed or confounded in some way. Results of studies such as these are not only meaningless, but also distort one's interpretation of the total pool of published research when these studies are included in a review. Based on their review, the ground rules provide a strong supportable set of conditions, however small, under which picture facilitation effects could be expected.
Even if you disagree with Levin and Lesgold's ground rules, at least you know what you are disagreeing with. For example, the first ground rule -- that passages should be presented orally -- makes generalization to other reading contexts arguable. However, the rationale of Levin and Lesgold to include this ground rule is simply to make certain that the subjects comprehended the context that the illustrations were supposed to support. This rule removed the possibility that any one subject's "inability to perform well on a prose-learning or comprehension test [could] be attributed to the subject's inability to read" (p. 234).
The importance of this issue cannot be understated when the goal is to interpret the effectiveness of illustrations on learning. Almost all of the available studies involve a research design in which the illustrations are supposed to support a passage of text. If subjects cannot read the text, then obviously there will be "no significant differences" (NSD) in varying the type of supporting illustration. The temptation is to conclude, incorrectly, that the illustrations had no effect. If subjects did not know how to read, then the results provide absolutely no information about the effectiveness of the illustrations. In fact, any time a study reports NSD, ineffectiveness of the instructional treatments is one of perhaps hundreds of rival hypotheses. Unfortunately, in all too many cases, the most probable cause of NSD is a problem with the research design. Levin and Lesgold go further by strongly suggesting that pictures will be of similar benefit when learners, when capable, are reading for comprehension.
By limiting their review only to studies involving children, as per the second ground rule, Levin and Lesgold remove the possibility that maturation effects may confound their interpretation. Similarly, the third ground rule, limiting the studies to those involving only fictional passages, greatly reduced the chance that subjects had prior knowledge of the information and therefore performed well on a test because they already knew the answers before the research began. This ground rule helps to "separate out what is learned from what is already known" (Levin & Lesgold, 1978, p. 235).
The fourth ground rule, that the pictures must overlap the story content, has remained one of the most fundamental principles of instructional graphic design. As discussed in chapter 2, pictures can serve a wide variety of purposes and functions, including decoration. It is important to distinguish these uses of pictures from those that try to illustrate a concrete feature of a passage. Levin and Lesgold's review provides evidence that pictures offer the potential to facilitate learning when the pictures are, indeed, congruent with the content to be learned.
The fifth and final ground rule is the most limiting for most practitioners. Levin and Lesgold limit their review only to studies in which factual recall was aptly measured. It would be inappropriate to generalize this finding to other types of learning, such as problem solving. Again, researchers frequently have been guilty of not only failing to precisely define the type of learning being studied in an experiment, but also failing to validate whether or not the testing procedures actually measured it. Due to the widely inconsistent definitions and measurements of learning in the available studies, Levin and Lesgold (1978) felt forced to limit their review to "factual information recall tapped by short-answer (generally "Wh") questions" (p. 236).
One example of the supportive research cited by Levin and Lesgold is a study by Guttmann, Levin, and Pressley (1977). In their study, children correctly responded to about 80% of the short-answer questions when pictures accompanied oral narratives, whereas children in a no-picture condition only answered about 57% correctly.
Taken as a group (you can not accept one condition without accepting the others), the ground rules from the Levin and Lesgold (1978) review provided the first meaningful set of guidelines to help practitioners predict when instructional visuals would be worth using. Further research continues to support these ground rules (see review by Levie & Lentz, 1982, for example).
Research Conducted and Reviewed by Francis Dwyer
One of the most prolific instructional visual researchers to date has been Francis Dwyer. Since 1965, Dwyer and his colleagues have conducted research with more than 8,000 high school students and 40,000 college students (Dwyer, 1972, 1978, 1987; Canelos, 1987). However, the most unique aspect to Dwyer's research is its systematic approach to investigating the use of visuals in instruction. The instructional materials of every study involved a 2,000-word script describing the parts, locations, and functions of the human heart. True to instructional design principles, instructional visuals were only added to illustrate parts of the script found difficult to learn. A total of 37 such critical areas were augmented with visuals (p. 52, Dwyer, 1978).
The types of visuals used in any one study depended on the research question being asked. Generally, the studies investigated the effects of representational pictures varying from highly detailed color photos to simple line drawings as an aid to learning the content contained in the script. Also studied were the effects of various lesson strategies on visualized instructions, such as using moving arrows and inserted questions as cueing strategies. A variety of treatment combinations were studied, such as incorporating the programmed materials into print-based materials, slide/tape materials with audio, and computer-based materials. This research effort, known as the Program of Systematic Evaluation (PSE), has produced more than 150 published research studies (Canelos, 1987).
In a way, Dwyer's research findings act as a testimonial to all of instructional visual research because they repeatedly show that visuals are not equally effective across learning situations. Effectiveness of all instructional strategies, such as visuals, depends on a wide array of factors, such as those already suggested by the research evaluation model in Figure 5.1. No true general principles were found. Some of the visuals were found to be effective some of the time under some of the conditions.
The most consistent results found by Dwyer were related to the amount of realism in the visuals. The results suggest that pictures facilitate learning for adults under certain conditions. For example, people need sufficient time to scan and interpret visuals with highly realistic details. Richly detailed visuals require the learner to attend to and systematically scan the visual in search of essential learning cues. If insufficient time is given, students may actually choose to ignore the visuals and attend to the more familiar, printed text. When lessons are externally paced, the most effective visuals are those that usually contain relatively small amounts of visual detail.
Dwyer's finding of the curvilinear relationship between learning and realism of instructional visuals (too little or too much realism adversely affects learning) is probably the hallmark of his research, and it disputed early realism theories (e.g., Carpenter, 1953; Dale, 1946; Morris, 1946) that suggested that the more information provided, the more learning would occur -- hence, more realistic pictures should invoke more learning. Dwyer's research indicates that when learners are confronted with visuals containing too much information and too little time they frequently choose to either ignore visuals or attend to wrong or inappropriate information in the visual. This phenomenon, where an optimal amount of arousal (not too little and not too much) is necessary for acceptable levels of performance, is consistent with many aspects of human nature, such as motivation, and is classically known as the Yerkes-Dodson law (1908).
Part of the value of Dwyer's research is to remind designers to be skeptical of any general, across-the-board recommendations to include visuals in instruction. Dwyer repeatedly reminds designers to weigh the effectiveness of materials against the efficiency and economy of producing them. If the instructional treatments containing visuals are no more effective than text alone, Dwyer (1978) encourages that visuals be omitted
Review by W. Howard Levie, 1987
Levie (1987) provides among the broadest reviews of picture research. (See Footnote 4) Levie reviewed available research in four areas: picture perception, memory for pictures, learning and cognition, and affective responses to pictures. Again, this review shows that much is known about too few of the "bricks" discussed in Figure 5.3. Levie suggests that "an aerial view of the picture research literature would look like a group of small tropical islands with only a few connecting bridges in between" (1987, p. 26).
The first two areas of inquiry, picture perception and memory for pictures, are essentially the same as those discussed in chapter 4. Research on recognition memory for pictures constitutes the largest pool of research on a single topic. This research is that associated with showing subjects hundreds of pictures and testing them: "Have you seen this picture before?" This research has obvious applications in advertising, for example, for issues such as product recognition. Among the important variables associated with a person remembering whether or not an image had been seen previously are meaningfulness and distinctiveness of the image, the ability to distinguish foreground objects from background graphics, complexity, color, movement, and the degree to which all objects in a picture have an obvious relationship to each other.
In considering memory for pictures, Levie, however, stresses the dual role of pictures: they are objects themselves, and, at the same time, they act as substitutes or surrogates for other objects. We need to remind ourselves that the interpretation of pictures is culturally based. For example, western cultures are inundated with information about the world represented in picture or iconic form. We tend to take for granted the degree of visual literacy of people brought up in those cultures versus other cultures. Some of these visual skills are probably innate, such as the ability to recognize and identify an object displayed in a picture. However, other visual literacy skills, such as decoding the meaning of abstract or complex visuals, require refined visual decoding skills.
The nature of what constitutes realism is also a very debatable issue. Again, we tend to take for granted the subjectiveness of picture recognition. For some, realism is somehow measured against the likeness of the object the picture is supposed to represent. For example, most people know that the object represented in a black and white photograph also contains color cues that are not presented. However, a second prominent view, one stressed in chapter 8, is that each of us constructs our own reality. In describing these two points of view, Levie (1987) writes that on one hand " . . . realism is basically a function of the degree of resemblance between the information provided by two types of optic rays" (p. 2), whereas, on the other hand, constructivists "point out that judgments of pictorial realism are influenced heavily by our learned preconceptions of how such a picture ought to look" (p. 4). This is just another reminder that perception, as discussed in chapter 4, is not like "taking a photograph" with our eyes.
The third area of research reviewed by Levie, learning and cognition, represents research on tasks more closely associated with school learning or training activities. This is the largest pool of research directly applicable to instructional design issues. All of the research reviewed and conducted by Samuels (1970), Levin and Lesgold (1978), and Dwyer (1972, 1978, 1987) fit here. Most of this research concerns the learning of verbal information, or facts. There is also ample research on the use of visual mnemonics as an aid to remembering information. For example, studies have shown that visual mnemonic techniques, such as the pegword system (see chapter 2), are generally ineffective in young children who are not able to internally image (Pressley & Levin, 1978). However, instructional techniques that provide a visual representation of the mnemonic have proven to be effective. Examples have included picture mnemonics that help children learn the names of minerals, state capitals, and U.S. presidents.
Little research has been done on using pictures in concept acquisition, and the research that has been conducted has not been systematic. As previously discussed, analogical pictures offer potential to help people understand abstract concepts (Tennyson & Park, 1980). Analogies that are visually based seem to facilitate learning by helping to make abstract information more concrete and salient (Mayer, 1983; Rigney & Lutz, 1975).
Levie notes that very little research is available on the role of pictures in higher-order thinking, such as problem-solving. However, there is some evidence to suggest that visual thinking plays an important part in skills associated with syllogistic reasoning (e.g., "If Tim is older than Mary and Mary is older than Brian, who is older, Tim or Brian?"). This research is not well-substantiated or corroborated; however, it is intriguing and seems to match many people's accounts of experiences of gaining insight while problem-solving. There are many stories of famous philosophers and scientists arriving at ingenious solutions to a problem by having the solution come to them in a vision. Einstein is said to have thought of the theory of relativity by imagining what it would be like to ride on a beam of light (Burke, 1985).
In contrast to research in the cognitive domain, little research is available on the affective effects of pictures. While there is little research to suggest that pictures necessarily promote stronger emotional feelings than words in the long-term, there is convincing evidence that pictures offer much more emotional impact in the short-term. One might refer to this as "shock value," and research suggests that pictures can arouse strong feelings in less than a second (Cupchik & Berlyne, 1979, as cited in Levie, 1987). Verbal forms of similar information, such as text, require comprehension and interpretation that can temper the emotional impact more than when presented visually. It has been suggested, for example, that television images of people being brutalized during the civil rights period of the 1960s helped sway public opinion and shape public action. There is certainly a difference between calmly reading an account of protesters being attacked by police dogs and actually seeing the animal's ferocity directed toward a 9- or 10-year-old girl or boy. Emotional reactions to films also have been documented (e.g., Lagerspetz & Englom, 1979 as cited in Levie, 1987).
Similarly, attitudes can be affected by pictorial information. Television news stories have been viewed as more accurate and believable when film shot on location is shown (Ryan, 1975). Also, research shows that people show preferences for types of visual material. For example, research shows that children often prefer realistic pictures. Other preferences include color, complexity, and ambiguity. However, one's aesthetic preferences are not necessarily related to times when the goal is learn something from the visual material. Dwyer (1972) has noted that "aesthetically pleasing visuals may be deceptive in their instructional value" (p. 90). Again, this serves as an important reminder from previous discussions that graphics can serve different functions, such as affective or cognitive. It is important that the intent of the graphics be defined. There is nothing wrong with adding visuals simply for aesthetic appeal, so long as these graphics do not undermine other instructional goals, such as those in the cognitive domain.
Review by Joel Levin, Gary Anglin, and Russell Carney, 1987
It should be evident by now that by far the most extensive body of literature related to the influence of pictures on learning has been that associated with prose learning. Prose learning essentially is learning from text. In contrast to the research cited earlier (e.g., Samuels, 1970), which studied the role of pictures in learning how to read, the emphasis in prose learning is on the ability to comprehend and remember information contained in the text, or "reading to learn" (Levin, Anglin, & Carney, 1987, p. 51).
Levin, Anglin, and Carney (1987) conducted a meta-analysis of pictures in prose studies. A meta-analysis is a statistical analysis that takes and groups the effect sizes of the independent variable in question (i.e., pictures) across studies. Using this pool of data, inferences about the independent variable can be made from the total group of studies. While open to criticism, meta-analyses offer the best procedure to date for objectively analyzing across studies. The result of their meta-analysis has yielded what they call the "10 commandments of picture facilitation." These commandments are presented here with "secular" translations to act as a concise set of guiding principles resulting from picture effects in prose learning (see Table 5.1). Together, these commandments and principles also act as a suitable summary and action plan for the research presented in this chapter.
Commandment 1: "Pictures shalt be judiciously applied to text, to remember it wholly" (p. 73). Instructional design principle: Since the available research is largely based on the effects of pictures on learning from prose, this research and any principles derived from it can only be consistently applied to similar situations.
This first commandment acts as an organizer for the rest. It serves as a reminder of where and how the commandments were derived. All of the remaining commandments and principles point to times when pictures offer the potential to facilitate learning from prose, such as text or narrative passages. This commandment also reminds us to not apply and generalize this group of research to a set of questions or problems that are inherently different. In other words, these principles are only directly useful if your goal is to design pictures and graphics to assist in remembering textual information presented in narrative fashion. If you are designing pictures or graphics for other tasks or functions, then you must look elsewhere for guiding principles. Unfortunately, most instructional designers deal with questions in which there is no substantive research literature. In those cases, a decision must be made whether or not research on the effects of pictures on learning from prose materials can be generalized to fit those circumstances. A pragmatic view is to cautiously accept this research pool until better and more direct information becomes available.
|
|
|
Commandment 2: "Pictures shalt honor the text" (p. 73). Instructional design principle: Pictures should be congruent and relevant to the information presented in the text.
This is among the most substantiated principles for the design of instructional graphics and is the same one promoted in the review by Levin and Lesgold (1978). Pictures must be congruent and relevant to the information or message contained in the prose material in order for the pictures to make a difference in learning. The most potent learning effects have been from the use of pictures that act as visual mnemonics. Further research shows that these picture effects are durable over time (Anglin, 1986, 1987).
Commandment 3: "Pictures shalt not bear false fitness to the text" (p. 73). Instructional design principle: If pictures are not congruent or relevant to the text, they may be distractive and interfere with learning.
In a real sense, this commandment is the complement to Commandment 2. As shown by research by Samuels (1967) and Willows (1978), pictures offer great potential for distraction. Obviously, when pictures are present and do not offer any instructional benefit, the risk that the learner will be distracted from the intended message is great enough to reconsider if the picture or graphic is necessary. This principle should be seriously considered by instructional designers and developers who seek to add graphics to increase the affective or motivational appeal of their materials. The use of cosmetic or decorative graphics offers the potential for two types of disruptive situations. The first is simply that appealing, cosmetic graphics may distract a learner's attention from the message. Given the fact that short-term memory is limited and temporary, any stimuli that offer the potential to occupy any cognitive processing should be carefully weighed. Second, a learner may actually be misled into believing that a graphic meant as decoration by the designer is somehow relevant to the message or instructional activity. This latter situation is particularly dangerous, as it opens the door for potential misconceptions.
Commandment 4: "Pictures shalt not be used in the presence of `heavenly' bodies of prose" (p. 74).
Commandment 5: "Pictures shalt not be used with text cravin' for images" (p. 74). Instructional design principle: Pictures will not be necessary when the text already prompts the learner to spontaneously form internal images.
These two commandments are presented together because they refer to situations in which pictures would not be expected to result in any greater learning than would be expected from the text alone. Commandment 4 refers to times when learners are already learning as much as possible without using additional aids, such as pictures. This is known in research circles as "ceiling effects." For example, if a learner already demonstrates mastery of a concept given one instructional method or strategy, then any additional strategy or aid will not result in more learning because there is no room for improvement. In regard to picture research, reaching a ceiling level of learning from text alone means either that the visualization of the content is not necessary to permit learning or that the text itself does a good enough job of cueing the learner to internally form an adequate image in short-term-memory.
Commandment 5 elaborates on the phenomenon, experienced by almost all people, in which the text or verbal message is so concrete and vivid that we are able to conjure up a suitable image in our minds ourselves. Probably the best examples of this phenomenon are from literature in which skillful writers "paint pictures" with their words. An example of this is contained in Box 5.1. Obviously, the ability of some words, whether written or spoken, to evoke mental images depends on a reader's or listener's background and prior knowledge. Of course, the ability to tell or write a story so that it paints pictures in a wide range of people's minds is a distinguishing characteristic of master writers and storytellers.
|
|
|
To think that you must actually draw a picture for people before they see something is often unnecessary (and perhaps presumptuous). Some of the best pictures are "drawn" with words. We all have our favorite examples, whether they be from writers or storytellers. One of my favorites is from John Steinbeck's (1939) The Grapes of Wrath which takes place during the "dust bowl" era of the 1930s. (I lived for a time along Route 66 in New Mexico, a road which figures prominently in the book.) Read the following passage and reflect on the images you "see" in your mind: Outside, the seated man stood up and looked over the cowl of the truck and watched the restaurant for a moment. Then he settled back on the running board, pulled a sack of tobacco and a book of papers from his side pocket. He rolled his cigarettes slowly and perfectly, studied it, smoothed it. At last he lighted it and pushed the burning match into the dust at his feet. The sun cut into the shade of the truck as noon approached. In the restaurant the truck driver paid his bill and put his two nickels' change in the slot machine. The whirling cylinders gave him no score. "They fix 'em so you can't win nothing, " he said to the waitress. And she replied, "Guy took the jackpot not two hours ago. Three-eighty he got. How soon you gonna be back by?" He held the screen door a little open. "Week-ten days," he said. "Got to make a run to Tulsa, an' I never get back soon as I think." She said crossly, "Don't let the flies in. Either go out or come in." "So long," he said, and pushed his way out. The screen door banged behind him. He stood in the sun, peeling the wrapper from a piece of gum. He was a heavy man, broad in the shoulders, thick in the stomach. He face was red and his blue eyes long and slitted from having squinted always at sharp light. He wore army trousers and high laced boots. Holding the stick of gum in front of his lips he called through the screen, "Well, don't do nothing you don't want me to hear about." The waitress was turned toward a mirror on the back wall. She grunted a reply. The truck driver gnawed down the stick of gum slowly, opening his jaws and lips wide with each bite. He shaped the gum in his mouth, rolled it under his tongue while he walked to the big red truck. The hitch-hiker stood up and looked across through the windows. "Could ya give me a lift, mister?" The driver looked quickly back at the restaurant for a second. "Didn't you see the No Riders sticker on the win' shield?" "Sure -- I seen it. But sometimes a guy'll be a good guy even if some rich bastard makes him carry a sticker." The driver, getting slowly into the truck, considered the parts of this answer. If he refused now, not only was he not a good guy, but he was forced to carry a sticker, and was not allowed to have company. If he took in the hitch-hiker he was automatically a good guy and also he was not one whom any rich bastard could kick around. He knew he was being trapped, but he couldn't see a way out. And he wanted to be a good guy. He glanced again at the restaurant. "Scrunch down on the running board till we get around the bend," he said. The hitch-hiker flopped down out of sight and clung to the door handle. The motor roared up for a moment, the gears clicked in, and the great truck moved away, first gear, second gear, third gear, and then a high whining pick-up and fourth gear. Under the clinging man the highway blurred dizzily by. It was a mile to the first turn in the road, then the truck slowed down. The hitch-hiker stood up, eased the door open, and slipped into the seat. The driver looked over at him, slitting his eyes, and he chewed as though thoughts and impressions were being sorted and arranged by his jaws before they were finally filed away in his brain. His eyes began at the new cap, moved down the news clothes to the new shoes. The hitch-hiker squirmed his back against the seat in comfort, took off his cap, and swabbed his sweating forehead and chin with it. "Thanks, buddy," he said. "My dogs was pooped out." Of course, if you've seen the film version starring Henry Fonda, your mental images may be clouded with this scene from the movie. In a sense, watching a movie before reading the book robs you of the experience of creating your own visual interpretation of the author's words. Instead, you are really watching another person's interpretation of the work, such as the film's director. One might argue, for example, that the symbolism of Melville's Captain Ahab in Moby Dick is too quickly replaced with the face of Gregory Peck, and, as a result, much of the symbolism is removed as well. The role and use of film versions of classic literature is an interesting issue to debate. |
From THE GRAPES OF WRATH by John Steinbeck. Copyright 1939, renewed ©1967 by John Steinbeck. Used by permission of Viking Penguin, a division of Penguin Books USA, Inc.
Commandment 6: "Pictures shalt not be prepared in vain" (p. 74).
Commandment 7: "Pictures shalt be faithfully created from generation to generation." Instructional design principle: When pictures are prepared as an aid to learning from text, be sure that the learner can read and understand the text.
The ability to read is an obvious prerequisite to experiments that investigate any aids to text comprehension. Unfortunately, the literature is filled with cases when this seemingly simple rule has been broken. Commandment 6 refers to the fact that pictures produced as an aid to learning from text cannot be effective unless learners have the prerequisite reading ability, such as word decoding and recognition, literal and inferential comprehension, and comprehension monitoring (E. Gagné, 1985). In these cases, the fact that pictures do not help "should be neither disturbing nor surprising" (Levin, Anglin, & Carney, 1987, p. 74).
Commandment 7 refers to the same issue, but is couched in the context of students being able to form internal images from text. Obviously, if students are not able to read, they consequently will not spontaneously form internal images of verbal information contained in the text. Other "sins" associated with commandment 7 include assuming that all learners are equally able to form internally images and that learners will be able to correctly decide which parts of the text should be used for internal imaging purposes. In other words, learners often have a difficult time distinguishing between important and unimportant text. As already mentioned, research has shown that not all people, such as young children, are appropriately able to internally image.
Commandment 8: "Pictures shalt not be adulterated" (p. 75). Instructional design principle: Pictures should be prepared to clearly represent the content that is to be remembered; additional "dressing up" of the picture should be avoided.
This commandment refers to specific graphic design decisions of the picture itself. For example, consider an instructional design that calls for a picture of a rose to supplement a text on gardening. Should a single rose be included, a bouquet, or the whole garden? Should line art or a photograph be included? Should color be included in the picture? The advice here is to avoid adding visual characteristics to graphics that do not directly support the clarity of the information to be learned. This commandment is supported largely by Dwyer's (1978, 1987) research on the realism of instructional visuals discussed earlier. Although countering instructional wisdom at times, this research suggests that too much information can often lead to poorer learning performance because learners often cannot extract the most relevant and salient information from a complex graphic.
Commandment 9: "Pictures shalt be appreciated for the art they art" (p. 75). Instructional design principle: The preceding commandments and related principles listed refer only to representational or analogical graphics, and not to arbitrary graphics, such as graphs and charts.
It would be a serious mistake to generalize research results taken from one class of pictures to pictures that serve totally different purposes and functions. It is important to remember that the previous commandments and related principles were generated from research where pictures were meant to bear some physical resemblance to objects, concepts, or processes represented directly or analogically to that included in a text.
Commandment 10: "Pictures shalt be made to perform their appropriate function" (p. 76). Instructional design principle: Pictures should be designed to perform their appropriate instructional functions based on the needs of the learner, the instructional objectives of the task, and the instructional materials actually used.
This commandment acts as a final summarization of the research on pictures as an aid to learning from prose. The amount of additional learning that can be expected by adding a picture is largely a function of the type of picture selected. The resulting instructional design principle summarizes this entire chapter, and in so doing, goes well beyond the commandment offered by Levin, Anglin, and Carney (1987). This principle suggests that the function of any particular graphic must be consistent with its intent. This principle also suggests that care must be taken to assure that the intent of a graphic does not subvert or undermine the function or effectiveness of other instructional elements, which may or may not be related to the graphic. For example, this principle accepts the premise that graphics can be added for affective considerations, such as those that serve cosmetic or motivational functions, but only under the condition that such graphics do not distract a learner's attention from any relevant information or tasks.
A FINAL WORD
The apparent contradictions concerning the effectiveness of pictures in reading require cautious interpretation. It is clear that there are contexts where pictures do not facilitate learning due to distraction effects and the inability of some readers to shift attention from pictures to text. However, ample contexts (the ground rules used by Levin and Lesgold [1978] for example) exist where pictures appear very useful in facilitating reading achievement. Dominant conclusions drawn from this research are: (1) pictures are superior to words for memory tasks; (2) adding pictures (external or internal) to prose learning facilitates learning, assuming that the pictures are congruent to the learning task; (3) children up to about the age of 9 or 10 rely more heavily on externally provided pictures than do older children; (4) children do not automatically or spontaneously form mental images when reading.
This chapter has tried to briefly present the research to date on the effectiveness of static visuals by "reviewing the reviews." This, however, is not a substitute for reading the reviews themselves and the primary sources of the research -- that is, the original published studies discussed in the reviews. Anytime one reads an author's interpretation of another's work, the resulting message becomes necessarily biased by the experiences and background of the reviewer. Therefore, it is important to remember that this chapter is but a beginning step in understanding the research on instructional uses of static visuals. (See Footnote 5)
Some of the next steps include reading these and other reviews, and as many of the original research reports cited as possible (and even those that were not cited). You should be in a position to list all the concerns and issues that influenced the reviewers' decision to include or exclude a particular study and also to decide if you agree with their decision. Finally, you must make your own interpretation and consider how the available research affects your instructional design decisions.
REVIEW