The Effects of Individual Differences in Spatial Visualization Ability

on Dual-Task Performance

The demands of the world today require us to be more and do more than ever before. Not only must we concentrate on our everyday needs, but we must now be able to do many things at once and do them well. What is it that enables us to perform certain tasks together? How is it that we can drive a car, maintaining good road position, while carrying on a conversation with our passenger? How do air traffic controllers keep an accurate mental image of their airspace while giving instructions to the pilots over headsets? Then, by the same token, why is it that there are tasks we simply cannot do well together -- that the attention needed for one task to be performed correctly, has a strong negative effect on the other task, such as carrying on two conversations at once, or having to monitor one visual display while tracking an aircraft on another display? Furthermore, what causes the great discrepancies in individual differences in terms of performance on these tasks? These are the questions and issues which will be addressed in this paper. This study will focus primarily on the issue of how people perform two concurrent tasks -- one of which is visual and spatial and the other, auditory and verbal. Situations, such as those described above as well as those which occur in today's electronic classrooms, often require us to take in auditory, verbal information, while at the same time maintain and process visual and spatial information. A scenario, relevant to the above examples, which occurs frequently in the electronic classroom environment, is one in which students must navigate through class notes on their computers, while listening to a lecture. This requires the students to use visual/spatial skills for navigation, while at the same time attend to the auditory/verbal lecture. Wickens' (1992) Multiple Resource Model will be used to understand the nature of this type of dual-task situation. Furthermore, spatial abilities, and the mental representation of images, will be investigated in an attempt to gain insight into how different people store and manipulate spatial information.

To begin, it would be helpful to look at the implications of dual-task performance -- how individuals allocate cognitive resources to different tasks, and under what conditions they can share these resources. To do this, we must first discuss Wickens' (1992) Multiple Resource Theory (MRT).

A central question to understanding how dual-task performance is possible lies in the area of resource allocation and limitations of these cognitive resources. While there are no clear-cut, agreed-upon definitions for cognitive resources, Wickens (1992) uses the concept of a channel, which describes the flow of information through the relevant processing stages and is "characterized by some distinct perceptual property as its source" (p.381). From this, he states that the term 'capacity' is used to "describe such items as the channel capacity of absolute judgment, the capacity of working memory, or the bandwidth capacity to transmit information along a channel in bits per unit of time" (p.381). Resources may then be thought of as bits information traveling along a particular channel, limited mainly by the capacity, in terms of size and speed, of that channel. In the past, it was believed that there was a single undifferentiated pool of resources and that these resources were available to all mental processing needs (Kahneman, 1973). This idea was based on Moray's (1967) concept in which, "attention was like the limited processing capacity of a general-purpose computer" (Wickens, 1992, p. 366). Tasks which were more difficult demanded more of these hypothetical resources than less demanding tasks, as did tasks which required a higher level of performance. Kahneman's extension of Moray's theory stated that as tasks become more demanding, thereby requiring more resources, physical manifestations, such as increased heart rate, sweating and pupil dilation would occur. This indicated that resources were being mobilized to be used for the demanding task.

What then happens when there are two tasks which need to be performed at the same time? We would assume that an increase in the performance of one task would result in the decrease in performance of the second task and vice versa. Norman & Bobrow (1975) developed the performance resource function (PRF), the hypothetical function relating performance quality and effort (in terms of resource allocation). If two tasks do interfere with each other, then an increase in the quality of the performance of one task results in the decrease in the quality of the performance of the second task due to shared resources. In terms of single-task performance, it maps a task against a "good" performance region (in which any effort leads to a high performance level) a "bad" performance region (in which no matter how much effort is exerted, the performance level is low), an "optimal" strategy (an increase in effort leads to a linear increase in performance level -- the harder you work, the better you will perform) and a "heuristic" strategy (a quick initial increase in performance, with little effort, but hitting an asymptote early on). The heuristic PRF describes much of human decision making effort -- we take shortcuts to get our answers, but this also incurs a risk since we often don't check out all of the alternatives. In terms of time sharing of two tasks the PRF can be used to show how the resources are "split" between the two tasks and how the performance output is affected. The degree of interference between these two tasks will depend, however, on the individual PRFs. If an individual task is either so trivial as to require virtually no resources (such as remembering your name), or so complex that even with all available resources it cannot be adequately performed (such as performing multiple complex mathematical calculations in your head) it is referred to as data-limited. However, if resources are divided (X % of the resources will go to task A and (100 - X)% will go to task B) and neither task is data-limited then there must be some trade-off in performance, in this case, the tasks are said to be resource-limited. It is this resource-limited condition that is of interest with regards to dual-task performance in terms of this particular study.

As appealing as this single-resource model may have sounded, there were severe limitations, the first of which was that it could not explain the results from dual-task interference studies. It is expected that if there are two or more tasks performed at the same level of performance, the task which is more difficult would require more resources. However, there are many examples in real life and in the literature that show that this is not always the case (Wickens, 1976, 1992; Payne, et al. 1994). Indeed, Hunt, Pellegrino & Yee (1989) showed that there was "an ability to coordinate information from several sources that is independent of the ability to process information from one of (these) sources alone." Therefore, there can even be an improvement in performance when dual-task processing occurs, in terms of spatial and verbal processing. Shah & Miyake (1996) report that their experiment "provides preliminary evidence for separate pools of cognitive resources for the two working memories (one for spatial thinking and the other for language processing)" (p. 12) further suggesting that tasks divided across the spatial/verbal dimension may be time-shared with minimal interference. However, this is not always the case, as there are many spatial and verbal tasks which do not time-share so well. Furthermore, two seemingly similar tasks may produce dissimilar results when performed concurrently with another task. For example, Payne, et al. (1994), studied the interference between speech intelligibility levels on an auditory memory search task, with concurrent visual tasks. The visual tasks were either (a) an unstable tracking task, (b) a spatial decision-making task, (c) a mathematical reasoning task, or (d) a probability monitoring task. They found that only the spatial task and the math task were affected by low speech intelligibility levels, whereas, the tracking task and the probability task were not. Why would some types of visual tasks cause interference with the auditory task while other tasks did not? In this case, the interference occurred when the auditory task was performed concurrently with the two tasks that required the processing of the information rather than just viewing information.

Wickens addresses these limitations in his Multiple Resource Model, which argues against the single undifferentiated pool of resources. Instead, he suggests that, "at one level, resources may be defined by three relatively simple dichotomous dimensions" (a) two of these are stage related (early and central processing versus late processing), (b) two are modality specific resources (auditory versus visual encoding), and (c) two are defined by processing codes (verbal versus spatial). If any two tasks need separate resources rather than common resources on any of these three dimensions, three things will happen: (a) there will be more efficient time sharing, (b) changes in the difficulty in one task will not have much of an effect on the performance of the other task, and (c) the performance operating characteristic (POC), which is constructed from two individual PRFs, will resemble two data-limited tasks because each task will need independent resources.

The main result of this model is that in some situations we can do two things at one time with very little decrement in performance to either task. However, this is only true if the two tasks occur either in separate stages, modalities, or codes. Some classic examples occur in the area of air traffic control. One such example in terms of stage-defined resources, is the ease with which an air traffic controller can communicate verbally with a pilot, while at the same time perform the spatial task of monitoring and maintaining a mental model of the airspace (Wickens, 1992). This ability to perform verbal/spatial dual-task processing is not only important in air traffic control, but also in driving (we can listen to directions (verbal), while we maintain a mental model of the road (spatial)), and can also be applied to the electronic classroom lecture environment, in which a student may listen to a lecture (verbal) while navigating through the three-dimensional space of the Internet or other related class programs, such as HyperCourseware™ which often involves manipulation of spatial images.

However, there are conditions where interference does occur between two tasks despite the predictions of MRT. In a study on communication between pilots and ground controllers, Loftus, Dark & Williams (1979) noted that under certain conditions, dealing with controller-issued instructions (an audio/verbal task) creates a heavy memory load for the pilot when the pilot is also performing an additional task, such as reading charts or checking the instruments (spatial/visual tasks). They noted that error rates for the pilots tend to increase when the complexity of the message is increased (e.g. when it contains more than one instruction). What is responsible for the differences in the results between these studies? One factor may have to with the different ways people perform spatial processing and the mechanisms of how we actually encode and process spatial information as well as the effects of individual differences. The way in which we represent mental images and the manner in which we can manipulate that information has long been a topic of debate (Tye, 1991).

In order to mentally explore and travel within our environment, we need to be able to represent visual and spatial images in our minds. What this means is, that we need to be able to store two- and three- dimensional information in a way such that we can later retrieve these representations and manipulate them as needed. For example, we represent images from maps so that they may later be used to help us navigate our way through new areas. The questions then, are, how do we encode these images and how do we process that information for practical use? Some theories state that pictures are encoded both verbally and imaginally, other theories address the representation of visual images as "quasi pictorial" formed from matrices of information, and still others assert that visual and spatial images are stored as propositional links. In this section, we will look at Paivio's (1979) Dual Coding Theory, which suggests different codes for storing concrete visual images versus verbal information. Then we will look at the "Imagery Debate" between the Pictorialists and the Descriptionalists, and discuss some methods of the lower level processing involved in the storage and transformation of the visual and spatial images.

Dual Coding Theory.

It was previously noted how the ability to encode verbal and visual information and to then integrate that information can be beneficial for learning and memorization. One further explanation for this ability to incorporate two codes comes from Alan Paivio's (1979) Dual Coding Theory (DCT). This theory makes the assumption that pictures and words activate independent imaginal and verbal codes and that the availability of each code differs. In general, he states that pictures and other visual objects are coded dually (as an image and verbally) so that there are two ways, to access the information. This would mean that, for example, a picture which would be coded using both an imaginal code and a verbal code would be more available than an abstract word which would be coded only using a verbal code (Paivio, 1979). Paivio updated his division of these two classes of phenomena in 1986, specifying that the verbal system refers primarily to language and the imagery system refers to nonverbal images, such as the analysis of scenes and generation of mental images. In addition, he emphasizes that while these two systems are functionally and structurally distinct, as well as functionally independent, they are interconnected so that they can act alone or in parallel. In this way, one system may trigger the other in a one-to-one or one-to-many bi-directional fashion. This ability to utilize two codes, would therefore, support the use of pictures and visually represented information in the educational setting and in terms of learning, in general.

Many studies have supported Paivio's Dual Coding Theory (Clark, 1987, Mayer & Anderson, 1991, Mayer & Anderson, 1992, Mayer & Sims, 1994) . Dickson, Schrankel, & Kulhavey (1988) determined from their study of verbal versus spatial encoding that, "Dual Coding is a constructive process in memory that results in conjoint retention of both spatial and linguistically referenced information" (p. 156), and that "two codes are better than one" (p. 156). Other studies point out the benefits of using pictures and words presented "contiguously in time or space" (Mayer & Anderson, 1991, p. 450), rather than just words alone, and note how the visual mode enhances verbal text and can often describe information in a way that is not possible with text alone (Steinberg, 1991). Studies in the area of television broadcast have also addressed this issue in terms of the redundancy between visual and auditory information (Drew & Grimes, 1987; Grimes, 1991; Lang, 1995; & Newhagen, 1995). Furthermore, Newhagen (1995) emphasizes how visual and auditory redundancy can enhance memory even when the auditory information is nonverbal.

However, it should be cautioned that audio and visual redundancy does not always lead to improved performance. Newhagen (1995) notes that when the audio and visual information is not in sync, or when two different tasks are involved it can cause an increased cognitive load. Gunter (1980) also warns that adding a visual context may not always lead to improved learning, that it is the "nature of the visualizations presented" (p.128) that is important. When the visual and auditory images have an easy association then they will be beneficial to each other. Otherwise, they may cause distraction and interference, such that the visual images might even inhibit learning of the auditory/verbal information. As shown by these studies, it is not just the simple matter of pairing pictures and words together that leads to enhanced memory, but rather pairing them based on the content and the compatibility of information they provide. In terms of the Dual Coding Theory, the high imagery of incoming pictorial information is effective because it increases the probability that both the imaginal and verbal codes will be activated. This is only beneficial, though, in terms of redundancy if that verbal code matches the information from the incoming auditory/verbal message.

It is clear then that the Dual Coding Theory addresses such issues as the independent activation of imaginal and verbal codes and the assumption that pictures and other visual objects are coded dually. What has not yet been addressed, is the nature of the code which is used to represent those visual images.

The Imagery Debate

Michael Tye (1991) discusses the imagery debate between Kosslyn's Picture Theory (1980) and Pylyshyn's (1973) anti-imagery stance. Both of these theories attempt to describe the hypothetical lower level processing involved in the encoding and transformation of visual and spatial images.

Kosslyn's Picture Theory: Kosslyn's theory states that even though mental images are not exactly spatial pictures as we see them -- they lack many spatial characteristics -- they still function as though they had those characteristics. Tye (1991) refers to these images as "functional pictures" (p. 35) and Kosslyn sometimes terms these "quasi-pictorial image representations" (Kosslyn, 1980, p. 30). In either case, this theory states that spatial images are represented as a matrix, with each entry in the matrix containing information, but that the information is really only meaningful when considered as part of the whole. In addition, certain properties of the spatial picture are preserved in the mental image. For example, each part of the representation must correspond to a portion of the actual object, and the relative distances within the portion are consistent with the actual object in terms of the number of locations between any two locations. In addition, Kosslyn's quasi-pictorial image representations allow for "processing that is sensitive to their internal representational structure (which) thereby enables the cognitive systems of which they are parts to treat them as if they were pictures with respect to such uses" (Tye, 1991, p.40). This processing involves "generating", "inspecting", and "transforming" images that reside in the visual buffer. We generate images from information in long-term memory and create an image in the visual buffer. We can inspect the image in order to recognize shapes and other characteristics of the object and make decisions based upon our inspection. Finally, we can perform transformations, such as rotation, folding, and scaling. An experiment which lends credence to this viewpoint involved mental rotation and was conducted by Steven Shwartz (1979). The most interesting and relevant result, in terms of this discussion, was that the greater the angle the experimental object was to be rotated, the longer it took to carry out the task. In addition, as the angle increased in arc length, large patterns required more time to rotate than small ones, but it did not make a difference whether the patterns were simple or complex. Likewise, Roger Shepard, in his studies of mental rotation demonstrated quite clearly a linear increase in response time as a function of degrees of rotation (Shepard & Metzler, 1971). Additionally, studies by Cooper & Shepard (1973) which examined the effects of prior information on mental rotation also lend support for this viewpoint. The time it took for the subjects to prepare for the upcoming image, given such advanced information as identity and orientation, was consistent with earlier findings. These findings suggest that "the images were not fading and being regenerated in different positions, but that subjects were rotating them by shifting them through the medium" (Kosslyn, 1983, pp. 108-109). These results have also been demonstrated in terms of size transformations and relationships (Kosslyn, 1983), and mental travel time as related to actual distances (Kosslyn, Ball, & Reiser, 1978). Findings from the latter study demonstrated that the amount of time taken on a "mental journey" was a function of the actual distances given on the map. These findings all lend support for a visual representation.

Pylyshyn's and Hinton's Descriptionalism: Although quasi-pictorial image representations are not exactly pictures they are definitely a different representation than the sentence or descriptionalist representation as offered by Pylyshyn (1973). The descriptionalist viewpoint suggests that "a structural description of an object is simply a complex linguistic representation whose basic non-logical semantic parts represent object parts, properties, and spatial relationships" (Tye, 1991, p. 61). Geoffrey Hinton, another descriptionalist, agrees with the basic premise of Pylyshyn's theory, but goes one step further to state that mental imagery is comprised of viewer-centered information with "object-centered structural descriptions of objects' shapes" (Tye, 1991, p. 71). Generating a mental image using Hinton's theory, would involve activating specific nodes in a hierarchical database stored in long-term memory. In addition, viewer-centered information would be attached to the general information from the active nodes. Another theory which utilizes the propositional approach is that of Clark & Chase (1972, 1974). They propose a theory which describes how people encode pictures and how they compare them to sentences. For example, they suggest that for the following picture: * the observer will first code the picture in an abstract propositional form, either as ABOVE(A,B) or as BELOW(B,A), where A and B represent the star and the line respectively. They then translate these codes into their English language surface representations -- "The star is above the line" or "The line is below the star" respectively. The findings from their studies suggest that:

(1) sentences are represented in terms of elementary propositions, (2) pictures are encoded in the same interpretive format, (3) these two codes are compared in an algorithmic series of metal operations, each of which contributes additively to the response latency; and (4) sentence encoding, picture encoding, comparing, and responding are four serially ordered stages, and their component latencies are additive (Clark & Chase, 1972, p. 472).

In general, whereas Kosslyn's Pictorial Representation appears to be more visually and spatially demanding in terms of resource usage, this propositional system seems to rely more on the use of verbal resources.

Both of these theories have their proponents and their detractors, however, the best theory may lie somewhere in between. Shah & Miyake (1996) note in their experiment on the separabiltiy of spatial and verbal resources, that some of the participants reported using a "variety of spatial and even verbal strategies to remember arrows in the simple arrow span task, ranging from integrating the arrows in a set to form a coherent figure ('chunking') to associating orientations with colors or sounds" (p.11). The difference in strategies may be due to the demands of task requirements, as noted by Tversky (1975), "Subjects are shown to verbally encode sentences for simultaneous comparison with pictures but pictorially encode sentences for later comparison with pictures. This is taken as further evidence that subjects adapt their encodings or representations of stimuli to demands of the task" (p. 405). Also Lohman (1989) suggests that there are two types of spatial knowledge -- that some knowledge is best represented by the quasi-pictorial method while other types of knowledge rely on more abstract, "proposition-based memory representations" (p. 346). But these differences may also be due to differences within the individual. Macleod, Hunt, & Mathews (1978) used a variation on the Clark & Chase ("star above line") paradigm and found that subjects adopted one of two different strategies, either a linguistic strategy, in which the internal representations of sentences and pictures are converted to propositional form, or a pictorial-spatial strategy in which those internal representations are converted to pictorial form. Furthermore, they found that, "psychometric measures confirmed a clear difference between the two groups in spatial ability but not in verbal ability. This difference was consistent with the hypothesized verification strategies; the subjects using the pictorial-spatial strategy demonstrated markedly higher spatial ability" and that "the choice of strategy is predictable from his psychometric measures of cognitive ability" (p. 493). From these studies, it would seem reasonable to ask whether people select strategies based on their particular individual abilities. Macleod, Hunt, & Mathews (1978) found that, in general, the individuals who were high in spatial ability tended to use the pictorial-spatial strategy while the individuals who were high in verbal ability tended to use the linguistic strategy. However, these were not necessarily mutually exclusive. Overall, low spatial ability subjects never used pictorial strategies, but high spatial ability subjects used both strategies on a relatively equal basis.

An additional factor is task complexity. Cooper (1982) found that her subjects used two extremely different strategies in solving figural tasks, one involved a parallel, holistic approach, while the other used a serial, analytic approach, and this was on a fairly simple task. The discrepancy in processing styles varied even more so when the tasks increased in complexity. The fact that different individuals use different techniques for performing the same tasks in terms of spatial representation suggests that there do exist individual differences. The next question, then, involves investigating a possible source of these individual differences. One such factor could involve Spatial Abilities

Spatial Ability (SA) has been defined as:

the ability to perceive spatial patterns or to maintain orientation with respect to objects in space (Ekstrom et al. 1976), is often cited as being a good predictor of computer-user performance (Egan 1988, Egan and Gomez, 1985), Gomez, et al. 1986, and Vicente et al. 1987). (Stanney and Salvendy, 1995, p. 1185). While most would agree that SA is indeed a component of general intelligence, there is still much debate about the quantity and descriptions of the dimensions of which it is comprised (Stumpf & Eliot, 1997). While some of the factor analytic studies have suggested as few as two major factors (McGee, 1979), others have suggested five (Carroll, 1993) or even ten (Lohman, Pellegrino, Alderton, & Regian, 1987). Stumpf & Eliot (1995) have looked at a large number of spatial tasks and have come up with a general factor of SA they refer to as k, which, though not identical to the general factor of intelligence, g , does show some overlap. Despite the finding of this general factor of SA, they too, do not believe that SA is determined by a single factor. Not surprisingly, there are many models which attempt to relate spatial tests to abilities. Eliot & Smith (1983) collected hundreds of spatial tests and classified them into 16 categories which were then grouped into a recognition and a manipulation division (Stumpf & Eliot, 1997). Lohman's hierarchical model involves analysis of data from 35 studies. The resulting structure maps spatial tests to processes based on speed and complexity, and has three "major" spatial factors as well as some "minor" factors. The major factors included Spatial Relations (SR), Spatial Orientation (SO), and Visualization (VZ). Furthermore, this model states that "Spatial ability may be defined as the ability to generate, retain, and manipulate abstract visual images. At the most basic level, spatial thinking is the ability to encode, remember, and transform, and match spatial stimuli" (Carroll, 1993, p. 305). It would seem that these definitions do capture the general nature of Spatial Ability. It will be important to realize then, that the results of this particular study will not sufficiently generalize to Spatial Ability. This study was designed primarily to address issues initiated by the work of Vicente, Hayes, & Williges (1987). By looking at measurements from various psychometric tests they found that Vocabulary and Spatial Visualization Ability were the best predictors of task performance on searching a hierarchical file system. These results have implications in terms of design issues for computer systems. It is because of this study we chose to specifically address the spatial factor of Spatial Visualization Ability (SVA).

As was the case with SA, SVA also has had a number of different definitions over time and within different studies. Carroll (1993) in chapter 8 of his text on "Human Cognitive Abilities" traces the definitions of SA and the various factors attributed to it. He cites French's, 1951 definition of Visualization (VZ) as "probably the ability to comprehend imaginary movements in 3-dimensional space or the ability to manipulate objects in imagination" (pp. 315-316). As suggested by Salthouse, Babcock, Skvronek, Mitchell, & Palmon, (1990), SVA is the "mental manipulation of spatial information to determine how a given spatial configuration would appear if portions of that configuration were to be rotated, folded, repositioned, or otherwise transformed" (p. 128). Ekstrom's ETS factor kit manual (Ekstrom, French, & Harmon, 1976) defines the factor Visualization (VZ) as "the ability to manipulate or transform the image of spatial patterns into other arrangements" (p. 173). It is further noted that "The visualization and spatial orientation factors are similar but visualization requires that the figure be mentally restructured into components for manipulation while the whole figure is manipulated in spatial orientation" (p. 173). Included under this factor are the following tests: VZ-1 form board test, VZ-2 paper-folding test, and VZ-3 surface development test. It is the VZ-2 paper-folding test which will be the measurement used for this study. It should be noted, however, that there are those who do not feel that the paper-folding test is really a measure of visualization and would, in fact suggest a different definition of visualization altogether (Eliot, 1987). However, for the purposes of this study, and in order to be consistent with the definitions and tests used by Vicente, Hayes, & Williges (1987) the definitions given by Ekstrom et al. (1976) will be the ones used from here on out.

Spatial Visualization Ability -- Applications

As mentioned previously, the way in which users visualize a system and its underlying structure may be due in a large part to their level of Spatial Visualization Ability (SVA), which has been shown to be related to an individual's ability to navigate through a hierarchical data base (Butler, 1990; Vicente, Hayes, & Williges, 1987). Individuals who are categorized as high SVA are often able to visualize three-dimensional space in an efficient and relatively accurate manner. Those who are low in SVA tend to use less efficient means to store and manipulate their visions of three-dimensional space. With the advent of large computer databases and menu hierarchies, SVA has also become important in terms of navigation within the realm of computers (Butler, 1990, Vicente, Hayes, & Williges, 1987) and it has been shown to be a great source of individual differences (Hegarty & Sims, 1994; Juhel, 1991). Lohman (1989) observed from interviews with subjects who had completed paper folding tests that, "all subjects rarely solve figural tasks in the same way... Some subjects solve items on such tests by generating mental images that they then transform holistically" (p.346). These people are categorized as high SVA individuals who are especially proficient at "generating, retaining, and transforming mental representations..." (p. 346). He then stated that other individuals, who are often categorized as low SVA individuals use less visual means to solve these problems. These subjects tend to use devices such as general reasoning skills, or external aids, although most subjects reported using more than one strategy, often switching back and forth among them.

Is it the case, as Lohman (1989) suggests, that high SVA individuals actually use different methods of transformation than low SVA individuals? Might they also encode that information differently, so as to make the transformation process easier? Also, in keeping with the questions about dual-task performance, how are the differences between high and low SVA individuals affected when a spatial transformation task is performed concurrently with a listening comprehension task? Are the differences amplified? Do the two tasks have a higher level of interference, and if so, what might this suggest? If the two tasks do interfere more for the low SVA individuals, this might suggest some type of verbal processing along with the encoding and transformation of spatial information. As mentioned previously in terms of Multiple Resource Theory, spatial tasks and verbal tasks should be able to be performed concurrently with minimal interference since they use different cognitive resources. However, if, for some individuals, spatial processing does require some of the verbal resources, this would indeed cause interference in the two tasks.

As computers become necessary to complete even some of our most routine tasks, it is the low SVA users who may be at a real disadvantage. Norman (1994b) notes that "technology may amplify individual differences" (p. 196) in that the user's proficiency with a system and system power combine multiplicatively. As computer power increases, not only does the overall level of performance of high ability individuals increase over the low ability individuals, but the variability among individuals increases as well. This means that the differences among individuals increase at a rapid rate as technology becomes more powerful. Studies have shown that SVA is a factor which is highly related to user proficiency. In the previously mentioned study by Vicente, Hayes, & Williges (1987), SVA and vocabulary were the only two predictors out of 21 which accounted for a significant portion of the variance (r2=.45). In fact, SVA was the single most influential predictor (r2=.33). Other studies have supported the link between SVA and computer performance. Research by Norman & Butler (1989b) found a correlation of -.38 between SVA and number of deviations from an optimal path search using a probability-driven search. Also, Butler (1990), found correlations of .30 between SVA and ability to find target information in a hierarchical database search. One reason why technology would seem to require SVA is that in order to use today's computer systems, the user needs to be able to navigate through complex systems, and the ability to manipulate, and navigate in two- and three- dimensional space is a major factor of SVA. With large databases of information, it's easy to get lost going from point A to point B. Networks have made this an even more intimidating experience by providing access to an overwhelming amount of information to search. The World Wide Web (WWW) is a prime example of such an environment. The WWW heavily uses Hypertext™, a card-based system which connects information by links to allow users to navigate the system, and while "the web" is very rich in information and affords users the opportunity to browse and follow non-linear streams of thought, it can be terribly intimidating for novice and low SVA users. Even in smaller, more controlled environments, the same issues of navigation apply. Low SVA individuals are just as prone to get lost, frustrated and confused by any program that does not explicitly and easily make its underlying structure and contingencies known. We need to determine how to design the technology so as to address the specific needs of this population, since they comprise the group which will be hindered the most by the increasingly complex navigational mazes.

If we believe, then, that a major factor impeding low SVA users is their ability to mentally represent and navigate through spatial information, what are some possible alternatives to alleviate these difficulties? This problem becomes magnified when we look at dual-task processing, where the added load of a verbal task may take resources away from the spatial task. This is especially true if a large portion of the verbal resources are being used in addition to spatial resources, which may be the case with low SVA users.

Apparency and Manipulability.

One of the handicaps created by the current large-scale systems, such as the WWW and other large databases of information, is that while they provide almost unlimited access to an ever-expanding array of information they are often constructed in such a way that navigation is impeded due to ineffective or misleading cues. That is, users are often not provided with the necessary and correct affordances for successful navigation. The term, affordances, as originated by Gibson and defined by Norman (1988) refers to, "the perceived and actual properties of the thing, primarily those fundamental properties that determine just how the thing could possibly be used" (p. 9). For example, a chair is for (afforded for) support and therefore affords sitting. Furthermore, Norman notes that:

Affordances provide strong clues to the operations of things...When affordances are taken advantage of, the user knows what to do just by looking: no picture, label, or instruction is required. Complex things may require explanation, but simple things should not. When simple things need pictures, labels or instructions, the design has failed. (p. 9)

In a well-designed system, the purpose of each of the components should be clear. For example, selection of links on the WWW should take you to the indicated destination, and selection of icons and menu items should carry out the functions they describe. However, this is not always the case. Sometimes, the users must first perform some other operation which is not necessarily described, before they can travel to the link or select the function. Sometimes the names of the links are just misleading, for example, does the key labeled "back" take the user back to the last screen the user was on, or to the previous screen in the hierarchy. Different systems answer this question in different ways. This may not seem like much of a problem to some users, but to those low SVA users who have difficulty navigating, it can seem like a tremendous obstacle. How do they find their way through a system where the paths may not be completely obvious and may at times, even be misleading (after all, not all designers are attempting to create user-friendly pages, and even well-intentioned designers may not use good judgment!) One solution is to attempt to make the underlying hidden relationships and paths apparent so as to "off-load the spatial processing of images from the user to the interface" (Norman, 1994b, p. 201). The user then can rearrange the image -- allowing him or her to manipulate the image externally. Norman & Butler (1989a; also reported in Norman, 1991, p. 312-313) looked at four levels of apparency (1) buttons only (No Apparency), (2) buttons plus all links (Non Contingent Apparency), (3) buttons plus all links to the goal (Goal Apparent) and (4) buttons with links from the start to the goal (Start/Goal Apparency) in order to see whether the added graphical information would help users with navigation through a hierarchical database of information. Their results showed that conditions in which they were given no additional external information (Conditions 1& 2) the users had to perform trial and error processing in order to reach the goal. But those subjects who were able to see the paths (Conditions 3 & 4) reached optimal performance levels quickly, thus "nullifying any differences in SVA" (Norman, 1994b,

p. 201). In addition, a study by Alonso & Norman (1996b and in press) replicated the findings in terms of the two extreme conditions (No Apparency and Goal Apparent) and found a definite improvement in performance with the apparent interface, particularly for the low SVA users.

In a second study by Alonso & Norman (1997 and in press), a model was proposed describing the characteristics of (a) the user, in terms of SVA and the tendency to internalize the path to the goal, and (b) the apparency of the interface, and relating them to how the users approach the task of "discovering" the underlying structure and relationships which describe the correct paths (see Figure 1). In the model the user's SVA is defined on a continuum from high to low. Likewise, interface apparency is defined on a continuum from high to low, such that low interface apparency implies that contingencies within the system are hidden from the user, and high interface apparency implies that these contingencies are revealed to the user. These user and interface characteristics will determine how the user progresses through the task.

Figure 1. Model of Path Processing.

Other methods for addressing these problems have been suggested and employed. Metaphors and analogies have been used to aid the user by providing a mental model of the system, and have been shown to be quite effective (Carroll & Mack, 1985). Graphical User Interfaces (GUIs) have also provided another successful interface solution by utilizing the spatial metaphor to create physical representations of these metaphors. The problem, though, in terms of this study, is that these solutions tend to differentially benefit high SVA users over low SVA users causing them to lag even farther behind (Butler, 1990) and providing the greatest improvement for those individuals who need it least. What we hope to do is provide some suggestions for those individuals who are at a disadvantage in this environment, due to their decreased visualization ability. It is the hope of this study to be able to offer some alternatives or at least research and development directions geared towards these users in order to provide options to make navigation easier and thereby accommodate for large variances in SVA. As mentioned earlier, interface apparency is one possible solution. Since some of the benefits of interface apparency have already been shown in present research based in single-task performance (Alonso & Norman, 1996b and in press), it is hoped that there would be even greater benefits when implemented in a dual-task situation.

Individual Differences in Spatial versus Verbal Processing. In the previous section, Spatial Visualization Ability (SVA) was discussed with regards to how some individuals are easily able to manipulate two- and three- dimensional objects in their head (central processing capability), while others are not able to do so. An interesting question is how does this fit in with Multiple Resource Theory (MRT)? We have discussed how some studies suggest that people vary in their methods for representing three-dimensional information (Cooper, 1982; Lohman, 1989; and Shah & Miyake, 1996). If people store and manipulate their spatial representations differently, that is if some use a holistic, or pictorial representation which relies on predominantly visual/spatial resources and others use a more propositional, descriptional representation which relies more on verbal resources, there maybe a difference in how they perform in a dual-task situation. If they encode and perform central processing functions using spatial resources, then one would assume that there is no decrement when performing concurrently, a verbal task with a spatial task (this would follow logically from MRT). But what happens if a person uses more verbal resources when encoding and performing central processing functions for a spatial image? That is, if the spatial image is stored and transformed as verbal propositional statements (or some other non-spatial representation) won't this then have an effect on time-sharing with a verbal task since they will both rely on the same set of verbal resources? That is, since the first task which was a spatial task, now utilizes more verbal resources, it has essentially become a second verbal task which then would interfere with processing of the actual verbal task. The question centers on how a person encodes and/or generates mental images and then how the image is manipulated and/or transformed.

In order to address the issue of how these incoming images are stored and processed, at least for the short term, it will be helpful to briefly discuss the role of Working Memory as it applies to spatial and verbal information.

Working Memory and its Effect on Verbal and Spatial Processing. Working memory is defined by Baddeley (1992) as:

a multi-component model controlled by a limited-capacity attentional system which we termed the Central Executive. This was supported by at least two active slave systems, the Articulatory or Phonological Loop that is responsible for maintaining and manipulating speech-based information and the Visuo-Spatial scratchpad or Sketchpad, which holds and manipulates Visuo-Spatial information. (Baddeley, 1992, p.8)

This model was used to address some problems in the earlier Atkinson-Shiffrin (1968) dichotomous view of memory, in particular, the notion of a unitary short term memory system (STM). In the Working Memory model, traces of speech-based information (actual speech, sub-vocal rehearsal, or sub-vocal naming of visual images) are held and processed in the Phonological Loop, while traces of spatial information are held and processed in the Visuo-Spatial Sketchpad, which itself may be further broken down into visual and spatial components (Baddeley, 1992, and Shah & Miyake, 1996). The Central Executive is responsible for attentional selection, and integration of information. In their application of the Working Memory capacity model to aging and discourse comprehension, Hasher & Zacks (1988) note that for most general capacity models the availability of cognitive resources, which vary with the requirements of each task, has an effect on cognitive functioning. Furthermore, as competition for these decreasing resources increases, processing takes higher priority than storage in working memory.

In terms of comprehension, Baddeley (1992) states that while the Phonological Loop does play some role in comprehension of complex sentences, patients with STM deficits show only minor comprehension difficulties. This would imply that either the Phonological Loop is primarily a backup or that these patients have sufficient capacity such that deficits do not affect the entire Loop. Hasher and Zacks (1988) focus more on the Central Executive and it's role in "orchestrating and enabling the multiple processes that co-occur to make skilled comprehension possible" (p.196). They feel that Working Memory is greatly involved in comprehension and further note that the goals of the listener will affect what information makes it into working memory.

So what then, does this say about the relationship between the capacity of verbal working memory and language comprehension abilities? Researchers have been trying to answer this question through tests which measure reading span (Daneman & Carpenter, 1980) and by considering composite measurements (Waters & Caplan, 1996). Based on results from their research, Waters & Caplan (1996) found that there was no single factor which could be used to measure this capacity. This, then might imply that there is "no central pool of verbal resources and that performance on sentence span tasks should be conceived of and analyzed in the same fashion as that on other dual tasks" (p.78). To this end, they make a claim for two distinct sets of resource pools, one of which "may be used for aspects of on-line psycholinguistic processing" (p. 78) and the other which "may be used for other verbally mediated tasks" (p. 78).

For the purpose of this study, a test of listening comprehension (derived from the Nelson-Denny test of reading comprehension, 1976) will be used to assess individual differences in terms of this skill only. It will not be used as a measurement of verbal Working Memory, but rather as a comparison between two tasks.

In terms of Working Memory and Spatial Visualization Ability (SVA) though, the primary goal of this study is to focus on the ways in which SVA is affected by individual differences. Salthouse, Babcock, Mitchell, Palmon, & Skovronek (1990) note that low SVA subjects may need more "workspace" than high SVA subjects for the same level and degree of processing. They suggest difficulty for low SVA subjects when there are storage and processing demands, because these demands may exceed capacity. Salthouse, Mitchell, Skovronek, & Babcock (1989) also mention reductions in working-memory capacity for increased task complexity with older adults. In addition, Shah & Miyake (1996) note that, "low spatial visualization ability participants may have (had) difficulty maintaining spatial representations while performing transformations, suggesting that the individual differences in spatial visualization ability may be accounted for, at least in part, by differences in spatial working memory capacity" (p. 7). If this is true, then it would be interesting to find out whether there is a difference between high and low SVA individuals in how they store and manipulate these spatial images.

It should be clear that the way in which we store and process visual, and specifically spatial, information may have a great effect on our day-to-day experiences. We often need to attend to two things at once -- maybe listening to directions while navigating a car through an unfamiliar section of town -- or maybe balancing a checkbook while watching television. In addition, the ability to perform dual-task processing is also quite essential in areas such as Air Traffic Control (ATC) as well as in a computerized classroom, such as the AT&T Teaching Theater. In order to illustrate these last two situations, consider the following examples. In the first, an Air Traffic Controller must communicate verbally with the pilot (using auditory and verbal resources) and at the same time monitor a visual/spatial display. According to the Multiple Resource Theory (MRT) this is not a problem since the auditory/verbal task uses different resources from the visual/spatial task. This assumes that the spatial information is being stored solely as spatial information. What if, however, the spatial information is also being stored as verbal information? Then, one might think that there would be interference resulting in a decrement in performance due to the need to share verbal resources. If low SVA users do need to access verbal resources in order to process spatial information, they would be the ones most affected in this dual-task scenario. A second example could involve students taking classes in an electronic classroom (for example, The AT&T Teaching Theater at the University of Maryland). The AT&T Teaching Theater is set up (at least at present) as a lecture environment which encourages the faculty to enhance the lectures with the use of computerized and multimedia equipment. We generally assume that the students attend to the lecture, which requires the use of auditory and verbal resources, and simultaneously work on their computers, which, for navigational tasks, relies on visual and spatial resources. Naturally, there will be some interference because there is also verbal information on the computers, but the main question here is, how heavily are the verbal resources used for each task and does that vary from student to student? That is, are some students more hampered by trying to represent some spatial information as verbal, thereby requiring more verbal resources and thus making the interference even worse? This might occur most in the area of navigation -- students trying to work their way through a series of displays, not understanding the underlying structure and contingencies. This problem of representation may cause some students, mainly those with low SVA, to get lost and therefore frustrated more often than those students with high SVA, who are better able to represent the images spatially. This would also cause more interference between the lecture (the primary task which would require auditory and verbal resources) and navigation through the computer screens (the secondary task which would rely primarily on visual and spatial resources for high SVA students, and on visual and verbal resources for low SVA students).

The Electronic Classroom

An electronic classroom is a learning environment which is set up in a similar physical manner to the "traditional" classroom, but with multimedia equipment available for lecture and classroom support. It provides many of the benefits found in traditional classrooms, such as the socialization aspects of "face-to-face, same place, same time" learning (Shneiderman, 1992), as well as the advantages of computerized technology. There are many electronic classrooms currently operating throughout the country and have generally been met with favorable reviews (Norman & Carter, 1992; Slatin, 1992). Often the electronic classroom provides each student with a desk top computer and the opportunity to interact with multimedia equipment during and integrated into a class lecture. In the electronic classroom the multimedia may be configured to have, among other things, a collection of computers and audio-visual equipment, often coordinated by control panels. This lets the instructor coordinate the various devices to present information to the students, thus providing more and varied learning tools. This also provides different modalities (visual and auditory, for example) in which to present material in the hope that students will better retain and recall that information (Paivio, 1979; Paivio, 1986; and Dickson, Shrankel & Kulhavey, 1988).

The AT&T Teaching Theater is an example of an electronic classroom currently operating at the University of Maryland, College Park (Shneiderman, Alavi, Norman, & Borkowski, 1994). The specific configuration of the Teaching Theater will be discussed later. The classroom has been operating for over five years and successfully utilizes the integration of lecture and multimedia presentations along with computerized technology so as to provide a rich environment for learning. As part of this multimedia package is a learning environment, called HyperCourseware™, designed by Kent Norman, specifically to take advantage of the layout and design of the electronic classroom.

HyperCourseware™ in the Electronic Classroom.

Created primarily as an educational environment to be used in the electronic classroom, HyperCourseware™ (HCW) provides the students with hands-on control of their lecture notes and an opportunity to perform real-time simulations. It has been used and periodically evaluated for the past five years at the University of Maryland, College Park, in the AT&T Teaching Theater (Lindwarm & Norman, 1993, and Norman & Carter, 1992). Norman (1994a) designed HyperCourseware™ so as to utilize the familiar structures in a traditional classroom (e.g. syllabus, class role, assignment lists) as metaphors for computer functions so that students would feel comfortable using the program during lectures. In general, HyperCourseware™ attempts to utilize that which is best about the Teaching Theater -- the ability to combine lectures and student participation in one package (Lindwarm & Norman, 1993). HyperCourseware™ is stackware, a card-based system written in Object Plus™ and is a "system of interlocking programs and files that serves as an infrastructure by creating tokens on a computer network that represent the familiar objects of instruction such as the syllabus, the class roll, lecture notes, exams and grade lists," (Norman & Lindwarm, 1993, p. 218). The syllabus section and related cards are generally used in addition to a classroom lecture and serve a similar, but expanded purpose as class notes. In addition to the textual information on each card, there are many simulations (referred to here as manipulations) which can be run in real-time during the lecture. Some examples of HyperCourseware™ screens are shown in Figure 2.

In a typical class lecture, the students follow along with the course material by paging through the cards on their computers. The students have the flexibility to go forward to upcoming cards, or review previously discussed cards at their own pace. Navigational aids are provided so that the students can get in step with the instructor (Norman, 1994a). This allows for the varying learning styles of each student -- exploring ahead, staying right with the instructor or maybe lagging behind, and catching up at their own pace.

In this study, the focus will be on the ability to navigate through such screens and interact with manipulations during lectures. Such manipulations are reminiscent of a chemistry lab where either the instructor performs experiments in front of the students or where the students perform their own experiments. However, HyperCourseware™ allows the students not only to perform their own experiments, but to repeat them as many times as they desire during class on the computer and without any mess to clean up! The ability to perform many iterations of the experiment can be very helpful. Students can see how the results may or may not change based on varying input values.

By studying HyperCourseware™ in the environment of the AT&T Teaching Theater, observations may be made as to how learning occurs in an electronic classroom. In terms of dual-task performance, the goal is to further understand the benefits of a multimodal learning environment, as well as how to assist those individuals who may be having trouble from "information overload" which may result from the enormous quantity of input information (Alonso, 1995, 1996a) .

Multiple Resource Theory and Spatial Visualization Ability (SVA) have previously been addressed separately, but how exactly do these two fit together? Based on the Wickens model (1984) this update, suggests that for two concurrent tasks -- one of which, the VZ-2, a test of SVA (a visual/spatial task), and the second of which, a listening comprehension exercise, (an auditory/verbal task) -- the high SVA and low SVA individuals access and utilize different pools of resources. For the listening task both high and low SVA individuals use the auditory resources and the verbal resources. For the spatial task, the high SVA individual primarily accesses the spatial resources and the visual resources, while using only a few of the verbal resources which means that there is very limited sharing of the verbal resources. However, for the same spatial task, the low SVA individual uses only a few of the spatial resources, primarily accessing the visual and verbal resources, which would mean that both the listening comprehension task and the VZ-2 task would involve heavy usage of the verbal resources. This, then, would lead to a large interference between the two tasks, such that one or both of the tasks would be performed at significantly less than the optimal level (see Figure 3).

Figure 3. Resource Usage during Dual-Task Performance of a Visual/Spatial task and an Auditory/Verbal Task, based on a model from Processing Resources in Attention by C. D. Wickens, 1984, in R. Parmasuraman & R. Davies (eds.), Varieties of Attention. New York: Academic Press. Thickness of lines corresponds to relative amount of processing impact on system.

The current study will focus primarily on the relationship between Wickens' (1992) Multiple Resource Theory Model and how the representation and manipulation of spatial information affects dual-task performance. In particular, this study will center on the following questions: Do individuals with low SVA process spatial information differently than high SVA individuals? Do low SVA individuals use more verbal resources than high SVA individuals? If they do not, then time sharing between a verbal and a spatial task should follow as predicted from Wickens' Multiple Resource Model for both high and low SVA individuals, however, if they do, then low SVA individuals should show more interference between the spatial and the verbal tasks. If their encoding and/or central processing uses spatial representations, then one would assume that there is no decrement when time sharing with a verbal task. But what happens if a person encodes and/or processes a spatial image as verbal propositional statements (or some other non-spatial representation)? This may have an effect on time sharing with a verbal task since the spatial task is now being treated as a visual/verbal task rather than a visual/spatial task.

In their experiment, Shah & Miyake (1996) came upon

an unexpected finding. They put forth one explanation for it as

"the possible use of verbal strategies, among some participants, to

encode spatial information in the span tasks" (p.20). They further suggested that, "if verbal strategies are used on some occasions to maintain spatial information then there may be interference from

verbal processing tasks as well as spatial processing tasks on the maintenance of spatial information" (pg. 20). From these statements, follows the primary hypothesis of interference for low SVA individuals performing dual-task processing involving a visual/spatial and an auditory/verbal task. If verbal strategies may be used during spatial processing, as suggested by Shah & Miyake (1996), and if low SVA individuals tend to be the ones who use more of these verbal strategies, as suggested by Lohman (1989), then the effects would be expected to show up as performance deficits, probably most noticeable in the spatial task, during dual-task processing, for those low SVA participants.

In order to look at how these individual differences (low versus high SVA) affect the method of encoding and/or processing and how that, in turn, affects the individual's dual-task performance in terms of concurrent spatial and verbal processing, two experiments and an observational classroom study will be run. The first of these experiments will address the dual-task question in a basic research setting and will utilize some of the properties of the dual-task methodology (see Ogden, Levine, & Eisner, 1979). It should be noted that this particular study does deviate from the classic methodology in that while most dual-task studies measure reaction times, this one instead measures the number of items answered correctly and number of items per trial. Students will first perform a single spatial task and a single listening comprehension task, so as to get baseline scores, and then the two tasks will be performed concurrently. The differences between the scores on the concurrent tasks and the baseline scores will be analyzed to see whether, neither, both, or one of the two tasks, was affected by having to perform them simultaneously . The second experiment will look at whether apparency (making explicit underlying contingencies so as not to overload the cognitive resources) helps alleviate the performance differences between the two groups and will also utilize some of the properties of the dual-task methodology. The observational classroom study will address the previous two questions but in a semester-long classroom setting in which both verbal and SVA resources are required in order to follow lecture material.

In general, it will be the goal of this research to (a) investigate how spatial information is mentally represented and manipulated, and whether there are important differences in how low and high SVA individuals store and process this information, (b) attempt to address the disparity between low and high SVA individuals on their performance in spatial tasks (and in turn on their performance in dual-task processing which involves spatial processing) by offering a strategy to alleviate some of the demands of spatial processing, and (c) observe students performing concurrent tasks (auditory/verbal and visual/spatial) in the "real world" context of a semester-long class in an electronic classroom. From these studies some methods or suggestions may be developed which can alleviate the anxiety and frustration toward dual-task processing, specifically for the low SVA individuals.

From the previously stated information, it is now possible to make the following four general hypotheses:

H1: High SVA individuals will show less of a decrement in performance from the single- to dual- task condition in terms of the visual/spatial task than low SVA individuals.

That is, the high SVA individuals will show more consistency between the single and dual tasks due to less interference between the auditory/verbal task and the visual/spatial task which in turn is due to less sharing of cognitive resources as based on the proposed model.

H2: Either, there will be no difference in performance level between the single- and dual- task conditions in terms of the auditory/verbal task, or the high SVA individuals will show less of a decrement in performance than the low SVA individuals.

This would be due to both the high SVA and low SVA using the same amount of cognitive resources as they had during the single-task phase of the experiment, although if there is sufficient need for the low SVA individuals to use verbal cognitive resources for processing of the spatial/visual task this would create interference and therefore a performance decrement.

H3: Low SVA individuals will perform worse than the high SVA individuals in terms of the verbal/spatial task in the non-apparent interface condition.

and

H4: Low SVA individuals in the apparent interface condition will perform at the same level as (or better than) the high SVA individuals in the apparent interface condition.

Apparency will alleviate any differences between the two groups such that the low SVA individuals will perform as well as this high SVA individuals in both the single- and the dual- task condition. Furthermore, both low and high SVA individuals are expected to perform at near optimal level from their first trial and will maintain that level throughout that phase of the experiment.

An additional hypothesis is:

H5: Either no correlation or negative correlation is expected to exist between SVA and Listening Comprehension Ability.

For example, it might seem that people who have high SVA might be lower in their listening comprehension ability, since they are more spatially oriented and not as verbally oriented. Likewise, it is also possible that low SVA individuals may have higher verbal abilities. Therefore, we would expect to see a negative correlation between the two groups in terms single-task performance, although this correlation may be rather small as this would probably be a rather weak effect. Additionally, there may be such a large degree of variability among individuals that no strong correlation would exist.

Finally, the following tables provide some predictions in terms of performance for the two experiments:

Table 1

In the single-task columns this table describes the predicted correlations between Spatial Visualization Ability (SVA) Level and each task, as well as Listening Comprehension (LC) Level and each task. In the dual-task columns, this table describes the changes in performance -- overall decrements and increases in performance. This suggests that the maximum effects of dual-task interference are expected to occur for the low SVA and high LC individuals. The high SVA individuals will probably not show much of a performance decrement, since there should be minimal task interference, unlike for the low LC individuals who will show a more substantial interference effect. Also, from hypothesis H5, it is assumed that there will probably not be any correlation, or possibly a negative correlation between LC ability and SVA.

In the single-task columns this table describes the predicted correlations between SVA Level and each task, as well as LC Level and each task. In the dual-task columns, this table describes the changes in performance -- overall decrements and increases in performance. This suggests that the major decrement in performance during the dual-task segment should occur for the low SVA individuals in the Non Apparent condition. High SVA individuals will probably have no problem in the Apparent condition, and may only have a slight decrement in performance in the Non Apparent condition -- that is, until they learn the underlying contingencies on their own. The low SVA subjects will take longer to learn the contingencies in the Non Apparent condition (that is, if they do indeed learn them). The low SVA subjects in the Apparent condition should perform at a level similar to the high SVA subjects, since with the assistance of the lines leading to the goal, they will not have to do much spatial processing.

This study consists of two experiments and one observational study, each of which examine how SVA affects the ability to simultaneously attend to spatial and verbal input. These experiments and the observational study were run in the AT&T Teaching Theater and the aITs classroom at the University of Maryland in College Park.

From the results of the two experiments and the observational study, it is the goal to gather information which would support and refine Wickens' (1992) Multiple Resource Model. Particularly, to understand how individual differences in SVA and comprehension fit into this model in terms of resource allocation and usage. Once the new model is better understood in these terms, solutions can be offered. One way to address this issue is with the use of apparency tools -- by making explicit the underlying structures and contingencies of the system. By alleviating the overhead of spatial visualization requirements for low SVA individuals they will be better able to devote more resources to the other tasks, such as attending to a class lecture. If it is recognized then, that technology may, in fact, have the ability to amplify individual differences and that low SVA individuals are the ones at the greatest disadvantage in an environment which emphasizes the use of technology, it becomes clear how important it will be to find solutions that address the needs of these individuals.

Next Page Back