Encyclopedia of Psychology
Contributor: Kent L. Norman
Article Title: Human-Computer Interface Design
The human-computer interface is the physical and conceptual boundary between the human user and the input/output devices of a computer. It is through this interface that the human gives instructions or supplies data to the computer and it is through this interface that the human receives feedback and other information from the computer. Most interfaces are dynamic and involve an interaction between the human and the computer.
Human-computer interaction (HCI) involves the activities of humans using computers. Interaction refers to a dialogue generated by the command and data input to the computer and the display output of the computer and the sensory/perceptual input to the human and motor response output of the human. Interaction takes place at the interface, which is made up of a set of hardware devices and software tools from the computer side and a system of sensory, motor, and cognitive processes from the human side.
Human/computer interaction is characterized as a dialogue or interchange between the human and the computer because the output of one serves as the input for the other in an exchange of actions and intentions. Early characterizations of the human-computer interface were more computer-centric. "The human-computer interface is easy to find in a gross wayjust follow a data path outward from the computers central processor until you stumble across a human being" (Card, Moran, & Newell, 1983). Today, the interface is viewed primarily from the other direction. Starting from human tasks and intentions, we follow the path of actions until we come to the machine.
The design of the human-computer interface requires a multidisciplinary approach. Research teams include psychologists, computer scientists, and specialists in subject matter domains such as business and management, library and information science, medicine, and so on. Although many of the developers and researchers in HCI reside in computer science and engineering, the contribution to HCI from psychology is key. Psychologists are interested in the perceptual, cognitive, and social aspects of the interface and tend to put the human before the computer, hence HCI. On the other hand, many computer scientists are concerned with the software and hardware that drives the interface and put the computer before the human, namely, CHI.
Psychologys contribution to interface design is many fold involving nearly all aspects of the discipline from sensation and perception of the computer screen and auditory output, learning and memory of commands and procedures, individual differences in experience and cognitive abilities using the computer and attitudes toward the computer, and developmental changes in appropriateness and usability of the computer.
Research in HCI has primarily focused on the cognitive processes involved on the part of the computer "user." From the perspective of the psychologist, the human-computer interface is a fertile testing ground for theories of cognitive processes. From the perspective of the designer, research in HCI helps to provide design guidelines and good design practices for interface development.
Historically, the field of HCI emerged from human factors on one side and computer science on the other. Although the distinction is blurred today, research in HCI tends to focus on computer systems while drawing heavily from the human factors literature, tradition, and methodology. Furthermore, work in HCI has expanded in specific applications such as database management, information retrieval, artificial intelligence, education, and multimedia.
A more detailed representation of the human-computer interface is shown in Figure 2 adapted from Norman (1991b). The human-computer interface is embedded within a task situation and environment. The user is performing some task such as monitoring operations in a factory or writing an email message. The interface is the overlap of areas representing the activities of the human (circle) and processes of the computer (square). The non-overlapping area of the circle represents the cognitive processes involved in tasks that are not directly focused on the human-computer interface. Similarly, the non-overlapping area in the rectangle represents the computer processes involved in tasks that are not directly related to the interface. From the perspective of the user, the overlapping area requires the mapping of intentions to activities such as keyboard entry and the mapping of screen displays to a meaningful interpretation of the information. From the side of the computer, the overlapping area involves the mapping of information in internal data structures to displays on the screen and the mapping of device input to internal codes.
Two basic flows of information and control are assumed through the interface: one going from left to right originating in the task environment directed toward the computer side of the interface, the other going from right to left originating in the machine environment directed toward the user side of the interface. It should be noted that at each point u-shaped arrows are used to indicate the feedback cycles of information flow through interfaces involving such processes as eye-hand coordination and error correcting.
Research in HCI has had a great impact on the development of new human-computer interfaces through guidelines documents, common interface specifications, and graphical interfaces. Most new software is subject to HCI guidelines and subjected to HCI usability testing procedures. Advances in computer ease of use and functionality are a direct result of research in HCI.
On the other side, cognitive psychology has been advanced by the study of HCI. Card, Moran and Newell point out the contribution of HCI to the current study of information processing in psychology. Models of users have been developed that track the information processing steps using computers. The use of metaphors and mental models of users has been researched. The impact of cognitive factors such as spatial visualization ability has been established. Research has contributed to our understanding of problem solving and the use of search strategies.
The importance of individual differences in the design of the human-computer interface has been emphasized in the rule, "Know thy user!" Users have been characterized along simple univariate scales from novice users to experienced users and along more complex cognitive and personality factors. Critical cognitive dimensions are reading comprehension and spatial visualization ability. Reading comprehension is particularly important in text-based interfaces and spatial visualization ability is important in graphical interfaces. Important personality factors that have been studied include (a) the need for closure, (b) whether one is field depend or independent, and (c) the degree that one is an introvert or an extrovert.
Reliance on Research Methods in Psychology
Studies in human/computer interaction rely heavily on established research methods developed in experimental and cognitive psychology. While a number of studies are observational or survey-based, the majority tend to use experimental methods. These involve defining an independent variable to manipulate in the interface. Generally, the independent variable pertains to some aspect of the human-computer interface such as the size, location, or images of icons on the screen, methods of inputting commands, aspects in the design of widgets, etc. The dependent variables include (a) performance variables such as task time and number of errors, (b) subjective assessments such as satisfaction with the interface, and (c) change variables measuring the time to learn a system and tendency to forget how to use a system.
Alternatively, the independent variable may involve the comparison of two different computer systems, programs, or versions. While such comparisons may indicate which of the two is superior, the results usually do not generalize or shed light on the reasons for the difference.
Models in Cognitive Psychology
The design of the human-computer interface is the result of a number of models. Figure 3 summarizes six models inherent in the interface.
The first is the interface model which is a top-level specification of the purpose of the interaction and the allocation of functions to the human and computer. For example, the human-computer interface might be a part of an air traffic control system. The purpose is to monitor and direct air traffic. Some functions are allocated to the air traffic controllers (e.g., selecting routes and communicating with pilots) and others to the computer (e.g., calculating distances and trajectories and alerting the controllers to potential traffic conflicts). Interface design must take into consideration both the top-level purpose and operation of the whole system.
The second model is the cognitive model of the operator generated by the cognitive psychologist. It is a model of the cognitive processes involved in performing a task. The cognitive model is a product of theoretical and empirical research on the limits, capacity, and information processing of the operator. The cognitive model helps (a) in the design of the interface by generating principals of good design (e.g., how to organize information, how fast to present information, how not to overload the working memory of the user) and (b) to make predictions about human performance.
The system model of the operator is a representation of the operators expected processing that is used by the system to predict operator behavior. The system model of the operator is used to predict or anticipate the actions of the human by the computer or disambiguate intentions of the user and format the output in a way to maximize operator comprehension and performance. The system may for example infer that when the user typed "ww.aps.org" he or she meant "www.aps.org" and automatically correct it.
The operator conceptual model is a representation of the system formulated by the designer and given to the operator to aid in the understanding and use of the system. This model is conveyed to the user through documentation, training, and the graphic design and behavior of the interface.
The operators mental model of the system is a representation within the mind of the operator of how the system works. The mental model is a collection of bits of declarative knowledge pertaining to the system and the task. For example, the users may think of a database as a library card catalog or think of a word processor as a glorified typewriter.
Finally, the interface object models are graphical or symbolic representations of token objects such as meters, gauges, knobs buttons, files and other devices represented on the screen. Interface object models help to engage the operators mental models of how things work or what things do.
It is important that these six models work in harmony and provide consistent matches between the elements and relationships among the models. Otherwise, unexpected events will occur; the users expectations will be incorrect; and errors will occur.
Interface Devices for Input and Output
The keyboard and pointing devices such as the mouse, trackball, and touch pad are the most common input devices for personal computers and computer workstations. The keyboard allows alphanumeric and cursor key input. Cursor positioning devices allow for continuous X,Y cursor positioning on the screen and clicks on screen locations. Touch screens can also be used for input by touching positions on the screen to make selections or dragging the finger across the screen to move objects on the screen. Voice input is possible using speech recognition hardware and software. Other more exotic devices include eye and head trackers, data gloves and bodysuits, and physiological readings. The design of input devices involves principles of motor control, feedback loops, and physiology from ergonomics and human factors psychology.
The computer screen, printer, and audio are the most common output devices. The computer printer allows for hardcopy text and graphic output. The screen provides transitory, dynamic output that can change as a function of state and that can show animation and video images. Audio output can provide the user with simple alerts, complex sounds, music, and speech synthesis. Other output devices include tactical stimulation to the body and stereoscopic images using virtual reality goggles. The design of the output devices involves principles from sensory and perception as well as cognitive psychology.
Modes of Interaction
The design of the human-computer interface requires methods for the transmission of information and control across the interface. Command languages allow the user to enter instructions via the keyboard. The computer system provides only a prompt indicating that the system is ready for a command. Command languages require the user to know the list of commands and the syntax for command statements. For example, if one wanted to copy a file from disk a to disk b the command might be: "copy a:file,b:file" More complex commands require programs. This mode of control is hard because it requires learning the commands, the proper syntax for generating statements, and the algorithms for complex programs. An extensive literature of the psychology of computer programming exists although it seems to have had only a minor effect on the development of easy to use programming languages.
On the other hand, interfaces that use menu selection do not require as much learning. Options are constantly available to the user and are structured to ensure proper logic. However, they do require the user to know or understand the functions of the options. Furthermore, with more complex systems involving hundreds or even thousands of menu items, navigation through hierarchies of menus is a problem. Good design of menu interfaces involves careful clustering, ordering, and labeling of menu options and the development of navigational tools to move through large sets of options such as those found on the World Wide Web.
Interfaces for data entry often use form fill-in as the mode of interaction. The computer screen shows labels for the types of data and text entry fields. The user can either use cursor keys or a mouse click to go the field and then use the keyboard to enter the data. Good interface design involves clear labeling of the fields and a logical and consistent layout of the items on the screen.
Finally, many "graphical user interfaces" (GUIs) use direct manipulation as a mode of input and control. Using direct manipulation the user can set values or perform functions by moving objects on the screen. For example, the volume of a sound can be set by using the mouse to drag a pointer along a volume control slider or the size of a circle in a drawing can be determined by dragging the perimeter out from the center. A file can be deleted by dragging its icon to a trash icon. Documents can be printed by dragging their icons to an icon of the printer. Good design for direct manipulation interfaces involves the study of eye-hand coordination, the meaningfulness of icons, an analysis of the screen layout and the steps that are required to perform a task.
Principles of Good Design
Although the term "user friendly" has been popularly used to characterize interface design, other factors have proven to be more practical and meaningful. For example, "ease of use" can be used to describe the ease with which a user can figure out how to perform a task and can then carry it out. Ease of use can be broken down into measurable units such as time to locate a desired function, the number of steps required to perform a function, the amount of time required for the task, and the amount of time needed to learn how to use the interface.
A number of design principles have been related to ease of use. For example, interfaces that are consistent in their use of terms, position of objects on the screen, and ways of doing things tend to be learned faster and are less subject to user errors. Interfaces that make good use of metaphors and create mental models are easier to learn and use because knowledge can transfer to aid in the understanding of the human-computer interface. Interface apparency increases ease of use by allowing the user to see visual representations of how things work rather than having to remember them or infer them. Immediate feedback and dynamic results allow users to see the consequences of their actions and make adjustments. Finally, undo-ability allows users to explore actions and easily correct errors.
Usability laboratories: Testing and Evaluation
Human-computer interface design is evaluated using controlled observation and testing. Many software and hardware companies do testing and evaluation in usability laboratories. Tests involve human participants who serve as computer users. The users are generally given a series of tasks to perform using the software (e.g., retrieve a record and change certain fields). The users are often video taped and encouraged to think aloud as they are performing tasks. Time to complete the tasks and accuracy are recorded by the testing software as well as complete records of all interactions between the user and the computer. In addition, users are asked to provide subjective assessments of overall satisfaction with the system, specific ratings of a number of characteristics of the system (e.g., helpfulness of the help system, meaningfulness of the terms used, difficulty finding desired options), and to suggest ways of improving the interface.
Ubiquity of the Human-Computer Interface
The human-computer interface is becoming more and more pervasive as the way of controlling and interacting with technology. The flight deck on airplanes has become known as the "glass cockpit" as knobs and dials have been replaced with human-computer interfaces. Home lighting and security systems use human-computer interfaces. Communication devices, automatic teller machines (ATMs), personal organizers, medical devices and monitors, educational toys, exercise machines, and so on all make use of the human-computer interface. Consequently, our understanding of the theories, principles, and optimal designs for the human-computer interface are extremely important and the contribution by psychologists is essential.
Card, S. K., Moran, T. P., and Newell, A. (1983). The psychology of human-computer interaction. Hillsdale, NJ: Lawrence Erlbaum Associates.
Norman, K. L. (1991a). Models of the mind and machine: Information flow and control between humans and computers. In M. C Yovits (ed.) Advances in computers. Volume 32. New York: Academic Press. Pp 201-254.
Norman, K. L. (1991b). The psychology of menu selection: Designing cognitive control of the human/computer interface. Norwood, NJ: Ablex Publishing Corp.
Norman, D., & Draper, S. W. (Eds.) (1986) User centered system design. Hillsdale, NJ: Erlbaum.
Shneiderman, B. ( 1998). Designing the user interface: Strategies for effective human-computer interaction (3rd ed.). Reading, MA: Addison-Wesley.
Author Name: Kent L. Norman