Judith S Olson. Berkshire Encyclopedia of Human-Computer Interaction. Editor: William Sims Bainbridge, Volume 2, Berkshire Publishing, 2004.
Human-computer interaction (HCI) is intimately intertwined with psychology. The field of HCI emerged as a special blend of human factors, cognitive psychology, engineering psychology, design, and computer science. The advent of generally available computing in the 1970s and the personal computer in the 1980s brought with them a concern that the interfaces were difficult to use, making people miserable and their work slow and frustrating. Psychologists were involved in human-factors research (research into designing equipment and environments for optimum human functioning) since World War II, so their interest in the problems of design in computing was a natural extension.
Psychology is the study of human behavior, ranging from the study of how neurons and physical structures determine behavior, to individual cognitive and emotional mechanisms, to the behavior of small groups and large organizations. Psychologists are interested in understanding both how computers are affecting people and how to design computers to fit better with human needs and capabilities. Because the computer is a very malleable and interactive tool, there has been a great interest in studying both its design and the effects of computer use on human behavior at all levels.
HCI And Organizational Psychology
Traditional HCI focuses on individuals and their interactions with the functionality offered through computer interfaces. Computer-supported cooperative work (CSCW) is an HCI specialty that focuses on group- and organizational-level issues.
For example, a key research finding at the organizational level is what is called the productivity paradox. Since the early 1990s, researchers have found zero correlation between the amount a company spent on information technology and the resulting productivity. Thomas Landauer, a psychologist, attributed this to poor design: The technologies are so difficult to learn and use that although they speed up some aspects of work, they slow down others. More recently Eric Brynjolfsson, a professor of management, has shown a small but positive correlation between money spent on information technology and productivity. Landauer would attribute that to an increase in user-centered design, which makes the technology easier to use.
In the late 1980s, the computer scientist Jonathan Grudin examined why people do not adopt various technologies that are intended to help them coordinate their work (like group calendars, for example). He found that many technologies benefited people other than the ones who had to expend the effort to enter information and keep it up to date. For example, a group calendar requires people to enter their meetings and appointments and keep them up to date so that others, such as their managers, can schedule more meetings. The cost is to the individual, but the benefit is to the manager. When calendars added features that benefited the person who had to expend the effort, they were suddenly widely adopted.
This issue of technology adoption ties in with issues of organizational change and change agents (technology being one such agent). Research may reveal remedies that facilitate technology adoption, either through suggesting changes in the design of the technology itself or in the design of the introduction of the technology to the people. There is a great deal known in organizational psychology that can be applied to the successful design and adoption of information technology.
Researchers in a related field, management information systems (MIS) in business schools, have researched important HCI-like topics, such as the effect of monitoring employee performance by computer (loss of control, the occasional lack of fit of the measure to the intended behavior) and the kinds of organizational structures that are helped most by various technologies.
HCI And Small-group Behavior
Social psychology has long studied group dynamics, trust, and impression formation, which are relevant to the study of how technology changes the behavior in groups. For example, there are a number of studies that show that people who communicate over video have little advantage over those with only audio in getting their work done. However, people prefer the video format to the audio-only format. Audio-only is hard. Without video, people lose valuable cues to turn taking and cannot discern whether their points are being understood or agreed with or not. People talk longer when they have no feedback about whether they are understood or not.
Trust is another relevant phenomenon that social psychologists have investigated. The old adage says “Trust needs touch.” Many people insist that a team needs to meet face-to-face in order to establish and maintain trust. Work in 2000 has shown, however, that if the team members engage in a social activity, even if they are not face-to-face, trust can and does form. It forms a bit more slowly, but in time it reaches the same levels as it would have had the team members met face-to-face.
Unfortunately, if a group contains both people who are collocated and people who are remote, those at a distance may be ignored and in some cases disadvantaged. Theories from social psychology suggest that this is because when dealing with people at a distance we lose the social cues, it takes more effort to communicate remotely, and any wrong step can cost a person membership in an “in-group.” Studies of telecommuting show that the principle of “out of sight, out of mind” applies. People who work remotely are passed over for promotion and not informed of important information.
The psychological concept of distributed cognition has similarly been embraced by those in HCI, especially those in CSCW. The core idea in distributed cognition is that the things that make us smart are in our environment; they include artifacts, other people, and the placement of both in space, not just in our heads. Studies of teams of pilots or ship controllers have shown that they construct cheat sheets and other aids and/or place team members in clear sight of each other to make their cognition easy at various stressful points. This recognition leads us to construct computer-based aids to team or individual performance, in particular looking at ways to use visual displays, for example, to help decision making or problem solving.
Researchers in HCI and CSCW have also worked with the concepts from psycholinguistics, primarily looking at the nature of conversations and the achievement of common ground. Common ground is people’s shared understanding of what they all know so that conversations include the right vocabulary and level of detail. Analysis of conversations with and without video, for example, show that it is easier to achieve common ground if you can see the visual cues of understanding or not. Without the visual cues of heads nodding or furrowed brows, people have to explicitly check with one another to see if they have understood or not.
Much of the work in social psychology relating to group dynamics has also been extended to the world of technology support for teamwork. The psychologist Joseph McGrath and his colleagues studied how groups form over time when they are not collocated and how moving people in and out of the collocated workspace affected trust buildup and people’s attention to one another and. Others have studied how different types of personalities affect the progress of groups, and that research too can be applied to the world of remote work. Researchers and practitioners in the business world are trying to assess which types of people would be good at working from home or telecommuting, and which would not.
In 2000, the HCI scholars Judy and Gary Olson published their findings on remote work, called “Distance Matters.” They highlighted the human aspects of teamwork that will never be solved with technology, such as cultural misunderstandings and circadian rhythms. The more we connect with people who are remote, the more likely we are to encounter people very different from us—the cultural distance will always be there. With large cultural differences, it is harder to achieve common ground and talk efficiently and effectively. Too much has to be explained. In addition, time zones and circadian rhythms cannot be overcome with technology. The Olsons noted that although when people travel they normally accommodate to the place visited, in videoconferencing, no one travels, and so less accommodation happens. Misunderstandings ensue.
HCI and Individual Psychology
In 1983, Stuart Card, Tom Moran and Alan Newell wrote a seminal book called The Psychology of Human Computer Interaction. In it, they compiled the findings about individual cognition that had to be attended to if computers were going to be easy to learn to use and then to use. This included facts about visual and auditory sensation, recognition, memory, problem solving, motor movement, and decision making.
In addition, they created a composite model called GOMS. GOMS has two parts: a knowledge base, and an engine that uses that knowledge base to act. The knowledge base consists of goals, operators, methods, and selection rules (hence the name GOMS); these four components specify what knowledge someone has to have in order to use a particular computer system. The engine, called the model human processor, uses parameters of human behavior (such as how long it takes to type, to point with a mouse, and so forth) in conjunction with this knowledge base and particulars of the environment (for example, the task to be performed) to predict behavior. The researchers also created a simplified version of this model called the keystroke-level model that helps predict how long it will take someone to do a task, taking into account both the physical (keystrokes) and mental (choosing between methods, retrieving a method) steps. In the mid-1990s, David Meyer and David Kieras added new components to the model to create EPIC (Executive Process, Interactive Control). EPIC focuses heavily on the perceptual motor aspects of using a computer and predicts well with complex multimodal tasks.
On a more practical level, many of the guidelines for designing computer systems and the methods for evaluating them rely on findings from cognitive psychology. Gestalt principles of perception (broadly, that people perceive things as unified wholes rather than as collections of parts) guide the layout of visual displays. Fitts’s Law, which concerns how rapidly a person can move from one point to another, guides keyboard design and the on-screen areas that can be selected or clicked on. An evaluation method called Cognitive Walkthough instructs the evaluator to ask questions about what users know how to do, whether they can find what they are looking for easily on the screen, whether they can perform the correct action easily, and whether they will understand what they just did. The method gets inside of the head of the users and assesses their knowledge, recognition, action, and understanding.
A recent addition to our understanding of how people behave with computers comes from a psychological construct of moment-by-moment analysis of the costs and benefits of doing various things. Stuart Card and Peter Pirolli have investigated and modeled people’s search behavior on the Web, noting how long they will continue with the results of a search, when they will quit and start a new search, and so forth. They call this behavior information foraging, with the analogy being to animals’ foraging for food. People make assessments of the likelihood that the next item will be good based on their assessment of the fit of the items up to that point, much the way a bird will continue in an orchard if he has success finding fruit. They use a concept called information scent to characterize that assessment.
The personality psychologist Sherry Turkle takes a very different approach to understanding the psychological aspects of human-computer interaction. Her books Life on the Screen: Identity in the Age of the Internet and The Second Self: Computers and the Human Spirit explore the feelings people have about their identity when taking a false persona in a chat room, or, in her case, finding that someone else was taking her identity. She also explores the thought patterns of the new generation of children, who refer to real life as RL and exclaim that some natural things look “realistic.”
Although not much has been done at the neural psychological level with computer interfaces, there have been a few intriguing experiments. People have been successful in manipulating a computer interface through their brain waves. The performance artist Stelarc went one step further, making it possible through a neural connection for someone to send a command to Stelarc’s muscle over the Internet, causing muscles to move. Stelarc has explored a number of other direct connections as well, including a robotic exoskeleton and a third ear and arm that are controlled by his brain.
Methods from Psychology
One of the earliest contributions psychologists made to the field of HCI was their set of methods. In HCI, we observe people at work to find out not just what they do at the computer but also how they organize their work setting. We observe them using the computer and note all the pausing, fumbling, going down wrong paths that occurs when they are having difficulty. In psychology this is called naturalistic observation.
Methods from experimental psychology have given computer designers the ability to evaluate various features, comparing people’s times and errors with various designs. Usability studies are similar to psychology experiments: Users are given a standardized task to do with a computer and monitored for their time, errors, and attitudes, just as subjects in traditional psychology experiments are give a task to perform and monitored as they perform it. HCI has adopted psychology’s rigorous standards for experimentation when it comes to testing both individual interface designs and general principles. For example, research into Fitts’s Law (which says that a user’s speed and accuracy selecting objects on a computer screen depends on the object’s size and how far the user has to move the pointer) has generated ideas about different kinds of computer menu designs, including circular menus and menus in which the further down the menu you go, the larger the area you click in.
Methods that ask people to sort items into categories have been used by computer designers to guide menu organization. Experiments have also been used to determine readability of fonts, color contrast, and other features of visual display. Additionally, experiments have helped determine how quickly things can move on the screen and still be successfully followed by the user. When windows open on the computer, they do not open instantly; you can see them get bigger—this is because the user sees the casual tie between their action (clicking on the file name) and the reaction (the enlarging window). Research also reveals what many of us know instinctively: that people are not very good at labeling files to be recognized by others or even by themselves at a later time.
What HCI Gives Back to Psychology
HCI is often thought of as an applied field, with psychology and other disciplines contributing to it. We believe, however, that in attempting to solve practical problems of design, HCI uncovers fundamental psychological phenomena, thereby contributing to both the practical and scientific world. Three big contributions HCI has made to psychology are integrated cognitive modeling (EPIC, inspired by GOMS), the theory of information foraging, and new methods for collecting data over the Internet.
As mentioned earlier, EPIC grew from the GOMS model, putting known component cognitive processes and stores together to predict complex behavior. In the spirit of other cognitive modeling efforts, the authors set a cognitive architecture and then program tasks using that architecture. A cognitive architecture is a set of assumptions about aspects of cognition that are relatively stable over time and relatively independent of the task. The architecture has an unlimited working memory in which information decays over time, a rehearsal process with particular time constants, separate perceptual and motor processors that have different time constants, and so forth. The model already has accounted for a wide variety of known phenomena in psychology. Because the model is programmed, it is easier to discover conflicting accounts of phenomena; the model makes thinking precise.
Information-foraging theory uses mathematically based ideas from biological food foraging to describe how people search for information. When some of the behaviors that Stuart Card and Peter Pirolli wanted to specify could not be easily put into mathematical terms, they turned to computer simulation. They also observed real people doing searches, both at their computers and in the library noting how information, books, stacks of paper, etc. were physically arranged. The theory predicted that a particular type of interface would support more effective searching, so Card and Pirolli built and tested it, confirming the design’s value. Their research is an excellent example of a multimethod investigation such as one rarely sees in psychology.
The last contribution is the design of Web-based experiments. It is now possible to collect data from thousands of people cheaply over the Internet. Questionnaires and surveys can be posted to sites such as SurveyMonkey.com, which will then send it out to a sample the researcher specifies and compile the results. Experiments can also be run online: Social psychologist Mahzarin Banaji, for example, has people download a small applet that then tests their reaction times to various pairs of words that can indicate people’s stereotypes and prejudices. More than 1.5 million responses have been collected at the site, a sample size far larger than any laboratory experiment could ever hope to collect.
Internet chat rooms are good places to observe interesting behavior—for example, researchers have been interested in seeing the reactions people get when they adopt a persona with a gender that is different from their own. Weblogs, or blogs, which are online journals or diaries, are also interesting sources of psychological data.
Psychologists and HCI in the Twenty-First Century
It is hard to think of any new technology that will not have psychological ramifications. In the twenty-first century, technology makes it possible for robots to be members of caretaker teams for the elderly; repair people can have the repair manual for the device they are repairing directly projected into their eyes by projectors mounted on glasses, the computer itself embedded in their clothing. Computers can sense tension in the user and suggest activities to reduce it. People can hear again thanks to cochlear implants—basically microchips that translate sound into neural pulses. As soon as children can reach a keyboard, there are educational and entertainment programs for them. People deeply care about robot pets. The World Wide Web is changing how we educate ourselves, both in the classroom and everyday life.
Sherry Turkle continues her study of identity and what it means to be human as the twenty-first century unfolds. For example, she is uncomfortable with an elderly person saying “I love you” to a robot and the robot saying “I love you” back. Although computers can be reactive in appropriate ways to people, she feels the emotion expressed in that instance is inauthentic, and thus troubling. Psychologists and people in HCI are likely to continue to be kept busy as the world embraces new information technology as fast as it is developed.