The Psychology of Human-Computer Interaction

Stuart K. Card, Thomas P. Moran, Allen Newell, Stuart K. Card

• Verlag: CRC Press
• Jahr: 2018
• ISBN:
  9781351409452
• Auflage: 1
• Thema: Informatique
• Thema: Interaction homme-machine

Defines the psychology of human-computer interaction, showing how to span the gap between science & application. Studies the behavior of users in interacting with computer systems.

1. An Applied Information-Processing Psychology

1.1. THE HUMAN-COMPUTER INTERFACE

1.2. THE ROLE OF PSYCHOLOGY

1.3. THE FORM OF AN APPLIED PSYCHOLOGY

1.4. THE YIELD FOR COGNITIVE PSYCHOLOGY

1.5. THE YIELD FOR COMPUTER SCIENCE

1.6. PREVIEW

A scientific psychology should not only help us to understand our own human nature, it should help us in our practical affairs. In educating our children, it should help us to design environments for learning. In building airplanes, it should help us to design for safety and efficiency. In staffing for complex jobs, it should help us to discover both the special skills required and those who might have them. And on and on. Given the breadth of environments we design for ourselves, there is no limit to the number of domains where we might expect a scientific knowledge of human nature to be of use.

The domain of concern to us, and the subject of this book, is how humans interact with computers. A scientific psychology should help us in arranging this interface so it is easy, efficient, error-free—even enjoyable.

Recent advances in cognitive psychology and related sciences lead us to the conclusion that knowledge of human cognitive behavior is sufficiently advanced to enable its applications in computer science and other practical domains. The years since World War II have been the occasion for an immense wave of new understandings and new techniques in which man has come to be viewed as an active processor of information. In the last decade or so, these understandings and techniques have engulfed the main areas of human experimental psychology¹: perception,² performance,³ memory,⁴ learning,⁵ problem solving,⁶ psycholinguistics.⁷ By now, cognitive psychology has come to be dominated by the information-processing viewpoint.

A major advance in understanding and technique brings with it, after some delay, an associated wave of applications for the new knowledge. Such a wave is about to break in psychology. The information-processing view will lead to a surge of new ways for making psychology relevant to our human needs. Already the concepts of information-processing psychology have been applied to legal eyewitness testimony⁸ and to the design of intelligence tests.⁹ And in the study of man-machine systems and engineering psychology, it has for some time been common to include a block diagram of the overall human information-processing system in the introductory chapter of textbooks,¹⁰ even though the reach of that block diagram into the text proper is still tenuous. There are already the beginnings of a subfield, for which various names (associating the topic in different ways) have been suggested: user sciences,¹¹ artificial psycholinguistics,¹² cognitive ergonomics,¹³ software psychology,¹⁴ user psychology,¹⁵ and cognitive engineering.¹⁶

Our own goal is to help create this wave of application: to help create an applied information-processing psychology. As with all applied science, this can only be done by working within some specific domain of application. For us, this domain is the human-computer interface. The application is no offhand choice for us, nor is the application dictated solely by its extrinsic importance. There is nothing that drives fundamental theory better than a good applied problem, and the cognitive engineering of the human-computer interface has all the markings of such a problem, both substantively and methodologically. Society is in the midst of transforming itself to use the power of computers throughout its entire fabric—wherever information is used—and that transformation depends critically on the quality of human-computer interaction. Moreover, the problem appears to have the right mixture of industrial application and symbol manipulation to make it a “real-world” problem and yet be within reasonable reach of an extended cognitive psychology. In addition, we have personal disciplinary commitments to computer science as well as to psychology.

This book reports on a program of research directed towards understanding human-computer interaction, with special reference to text-editing systems. The program was undertaken as an initial step towards the applied information-processing psychology we seek. Before outlining individual studies, it is appropriate to sketch how this effort fits in with the larger endeavor.

1.1. THE HUMAN-COMPUTER INTERFACE

The human-computer interface is easy to find in a gross way—just follow a data path outward from the computer’s central processor until you stumble across a human being (Figure 1.1). Identifying its boundaries is a little more subtle. The key notion, perhaps, is that the user and the computer engage in a communicative dialogue whose purpose is the accomplishment of some task. It can be termed a dialogue because both the computer and the user have access to the stream of symbols flowing back and forth to accomplish the communication; each can interrupt, query, and correct the communication at various points in the process. All the mechanisms used in this dialogue constitute the interface: the physical devices, such as keyboards and displays, as well as computer’s programs for controlling the interaction.

At any point in the history of computer technology there seems to be a prototypical user interface. A few years ago it was the teletypewriter; currently it is the alphanumeric video-terminal. But the actual diversity is now much greater. All so-called “remote entry” devices count as interfaces; and a large number of such specialized devices exist in the commercial and industrial world to record sales, maintain inventory records, or control industrial processes. Almost all such devices are fashioned from the same basic sorts of components (keyboards, buttons, video displays, printers) and connect to the same sorts of information-processing mechanisms (disks, channels, interrupt service routines).

The very existence of the direct human-computer interface is itself an emergent event in the development of computers. If we go back twenty years, the dominant scheme for entering information into a computer consisted of a trio of people. First there was the user, someone who wanted to accomplish some task with the aid of the computer. The user encoded what he wanted onto a coding sheet, then sent it to a second person, the keypunch operator, who used an off-line device, the keypunch, to create a deck of punched cards that encoded the same information in a different form. The cards in turn went to a third person, the computer-operator, who entered the cards into the computer via the card reader. The computer then responded by printing messages and data on paper for the operator to gather up and send back to the user. The relationship between the user and the computer was sufficiently remote that it should be likened more to a literary correspondence than to a conversational dialogue. It is the general demise of such arrangements involving human intermediaries, and the resultant coupling of the user directly to the computer, that has given rise to the contemporary human-computer interface. Whatever continued evolution the interface takes—and it will be substantial—human-computer interaction is unlikely ever to lose this character of a conversational dialogue.

Figure 1.1. The human-computer interface.

Of course, there is much more to improving computer interfaces than simply making them conversational. Informal evidence from the direct experience of users provides numerous examples of current interface deficiencies:

In one text-editing system, typing the word edit while in command mode would cause the system to select everything, delete everything, and then insert the letter t (this last making it impossible to use the system Undo command to recover the deleted text because only the last command could be undone).

In another text-editing system, so many short commands were defined that almost any typing error would cause some disaster to happen. For example, accidentally typing CONTROL-E would cause the printer to be captured by the user. Since no indication of this event was given, no other users would be permitted to print until the other users eventually discovered who had the printer. In an even more spectacular instance, accidentally typing CONTROL-Z would delete all the user’s files—permanently.

In one interactive programming system, misspelling a variable name containing hyphens (a common way of marking off parts of a name) would cause the system to rewrite the user’s program, inserting code to subtract the parts of the name. In many cases, the user would have to mend his program by hand, laboriously searching for and editing the damaged code.

In a set of different subsystems meant to be used together, the name “List” was given to many different commands, each having a different meaning: (1) send a file to the printer to make a hardcopy, (2) show the directory of files on the display, (3) show the content of a file on the display, (4) copy the workspace to a file, (5) create a particular kind of data structure.

Yet, when one looks at the teletype interfaces of yesterday, it is clear that substantial progress has been made. The emergence of the direct human interface, circumventing the keypuncher and operator, must itself be counted as an improvement of enormous value. We now have interfaces that allow the use of computers for such highly interactive tasks as making engineering drawings and taking airline reservations. But despite considerable advancements, the systems we have are often ragged and in places are sufficiently poor to cripple whole ranges of use.

What strikes one most noticeably about existing interfaces, besides all the little ways they fail, is that their failures appear to be unnecessary. Why, when interaction could be so smooth, even elegant, is it often so rough, even hazardous? Two observations may help explain this perplexing state of affairs.

First, interaction with computers is just emerging as a human activity. Prior styles of interaction between people and machines—such as driver and automobile, secretary and typewriter, or operator and control room—are all extremely lean: there is a limited range of tasks to be accomplished and a narrow range of means (wheels, levers, and knobs) for accomplishing them. The notion of the operator of a machine arose out of this context. But the user is not an operator. He does not operate the computer, he communicates with it to accomplish a task. Thus, we are creating a new arena of human action: communication with machines rather than operation of machines. What the nature of this arena is like we hardly yet know. We must expect the first systems that explore the arena to be fragmentary and uneven.

Second, the radical increase in both the computer’s power and its performance/cost ratio has meant that an increasing amount of computational resources have become available to be spent on the human-computer interface itself, rather than on purely computational tasks. This increase of deployable resources exacerbates the novelty of the area, since entirely new styles of interaction become available coincidentally with an increased amount of computational ability available per interaction. These new styles often lead to completely new interfaces, which are then even more ragged than before. At the same time, opportunities for the invention of good interfaces also increase rapidly, accounting for the leaps and bounds we have seen in terms of major improvements in functionality and ease of use.

1.2. THE ROLE OF PSYCHOLOGY

Many in the computer field agree that there is an obvious way to design better human-computer interfaces. Unfortunately, they disagree on what it is. It is obvious to some that psychological knowledge should be applied. Their slogan might be, in the words of Hansen (1971): “Know the user!” It is obvious to others that the interface should simply be designed with more care—that if designers were given the goal of good interfaces, rather than stringent cost limits or tight deadlines, then they would produce good designs. Their slogan might be: “Designers are users too—just give them the time and freedom to design it right!” And it is obvious to others still that one should pour the effort into some new components—flat displays, color graphics, or dynamically codeable microprocessors in the terminal. Their slogan might be: “Make the components good enough and the system will take care of itself!”

Who is to gainsay each of these their point? The technology limits, often severely, what can be done. All the human engineering in the world will not turn a 10-character-per-second teletypewriter into a high-resolution graphics terminal. The history of terminal development so far is writ largely in terms of advances in basic interface components, most notably the resources to allow substantial computational cycles to be devoted to the interface. It is easy to point to current limitations whose lifting will improve the interface by orders of magnitude. Immense gains will occur when the display holds not the common 24 × 80 characters (the typical alphanumeric video terminal, widely available today), but a full page of 64 × 120 characters (the typical 1000 × 800 pixel video terminal, available at a few places today), or even the full drafting board of 512 × 512 characters (not really available anywhere, yet, as far as we know).

Moreover, any accounting will have to credit the majority of the capabilities and advances at the interface to design engineers and only a few of them to psychologists. However many imperfections there remain in the interface, the basic capabilities and inspired creations that do exist came out of an engineering analysis of the functions needed and the fact that the designer, being human, could empathize directly with the user.

And yet, there remain the mini-horror stories—of systems where, after the fact, it became clear that either the nature or the limitations of the user were not appreciated, and some design foolishness was committed. Since it is these stories that come to mind in discussing the role of the human at the interface, it is often assumed that all that one needs are ways of checking to be sure that the obvious is not overlooked; “All we need from psychology is a few good checklists!” might be the slogan here. But as we shall see, there is more to human-computer interaction than can be caught with checklists.

The role psychology might be expected to play in the design of the user-computer interface is suggested by the results it was able to achieve for military equipment during World War II. At that time, it had become apparent that a strong limiting factor in realizing the potential of man-machine systems, such as radar sets and military aircraft, lay in the diffic...

Inhaltsverzeichnis

• Cover
• Half Title
• Title Page
• Copyright Page
• Dedication
• Table of Contents
• Preface
• 1. An Applied Information-Processing Psychology
• SCIENCE BASE
• TEXT-EDITING
• ENGINEERING MODELS
• EXTENSIONS AND GENERALIZATIONS
• Symbol Glossary
• Bibliographic Index
• Subject Index