Understanding Human-Computer Interaction For Information Systems Design

Understanding Human-Computer Interaction for Information Systems Design
Author(s): James H. Gerlach and Feng-Yang Kuo

Source: MIS Quarterly, Vol. 15, No. 4 (Dec., 1991), pp. 527-549
Published by: Management Information Systems Research Center, University of Minnesota
Stable URL: https://www.jstor.org/stable/249456
Accessed: 12-08-2019 01:48 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Management Information Systems Research Center, University of Minnesota is

collaborating with JSTOR to digitize, preserve and extend access to MIS Quarterly
This content downloaded from 103.99.43.12 on Mon, 12 Aug 2019 01:48:44 UTC
All use subject to https://about.jstor.org/terms
Human-Computer Interaction
design is divided into three major divisions:

Understanding Human- system model, action language, and presenta-
Computer Interaction tion language. The system model is a concep-

tual depiction of system objects and functions.
for Information The basic premise is that the selection of a good
system model provides direction for designing ac-
Systems Design tion and presentation languages that determine
the system's look and feel. Major design recom-
mendations in each division are identified along
with current research trends and future research
By: James H. Gerlach
Graduate School of Business issues.
Administration
Keywords: User-computer interface, user men-
University of Colorado at Denver
tal model, human factors, system
Campus Box 165 model, presentation language, action
P.O. Box 173364
language
Denver, Colorado 80217-3364
ACM Categories: D.2.2, H.1.2, K.6.1
Feng-Yang Kuo
Graduate School of Business
Administration Introduction
University of Colorado at Denver The user is often placed in the position of
Campus Box 165 an absolute master over an awesomely
P.O. Box 173364 powerful slave, who speaks a strange and
Denver, Colorado 80217-3364 painfully awkward tongue, whose obe-
dience is immediate and complete but
woefully thoughtless, without regard to the
Abstract potential destruction of its master's things,
rigid to the point of being psychotic, lack-
Over the past 35 years, information technology
ing sense, memory, compassion, and-
has permeated every business activity. This
worst of all-obvious consistency (Miller
growing use of information technology promised
and Thomas 1977, p. 512).
an unprecedented increase in end-user produc-
tivity. Yet this promise is unfulfilled, due primari- The problems of human-computer interaction
ly to a lack of understanding of end-user (HCI), such as cryptic error messages and incon-
behavior. End-user productivity is tied directly to sistent command syntax, are well-documented
functionality and ease of learning and use. Fur- (Carroll, 1982; Lewis and Anderson, 1985;
thermore, system designers lack the necessary Nickerson, 1981) and trace back to the beginning
guidance and tools to apply effectively what is of the computer revolution (Grudin, 1990). The
known about human-computer interaction (HCI) impact of problematic HCI designs is magnified
during systems design. Software developers greatly by the advent of desk top computers,
need to expand their focus beyond functional re- employed mainly by professionals for enhancing
quirements to include the behavioral needs of their work productivity. A faulty HCI design traps
users. Only when system functions fit actual work the user in unintended and mystifying cir-
and the system is easy to learn and use will the cumstances. Consequently, the user may not
system be adopted by office workers and adopt the system in his or her work because
business professionals. learning and using the system are too difficult and
time-consuming; the business loses its invest-
The large, interdisciplinary body of research
ment in the system.
literature suggest HCI's importance as well as its
complexity. This article is the product of an ex- As concern about HCI problems grew, research
tensive effort to integrate the diverse body of HCI was conducted by both practitioners and scholars
literature into a comprehensible framework that to find solutions. Initially, researchers focused on
provides guidance to system designers. HCI enhancing programming environments in order
MIS Quarterly/December 1991 527
to improve programmers' productivity. With the Green, 1986) are examples of HCI theoretic
proliferation of desk-top computers, it was models for studying user behavior (to be dis-
discovered that non-technical users were not cussed later). These models provide a basis for
satisfied with the same type of environment that
explaining why some design guidelines work. Our
programmers used. Research has since expand- second objective is to elaborate existing
ed beyond technical considerations to in- guidelines with their task constraints and
vestigating behavioral issues involving human theoretic bases so a designer can relate them to
motor skills, perception, and cognition for new, untested situations.
developing functional, usable, and learnable soft-
Our third and last objective is to identify oppor-
ware. HCI is now an important scientific discipline
tunities for HCI research. An exhaustive review
built upon computer science, ergonomics,
of guidelines and theories in user interface design
linguistics, psychology, and social science.
reveals gaps in our knowledge regarding the im-
Today's system designers are expected to apply pact of design choices on human behavior. By
these interdisciplinary principles to improve user noting these opportunities, we hope to interest
satisfaction and productivity. This is a formidable both practitioners and research scholars in fur-
task because HCI development is not an aspect thering our knowledge of user interface design.
of software design that can be illuminated by a
We begin with a framework for organizing HCI
single design approach. More importantly, there
design and several theoretic approaches to in-
is a lack of guidance in applying HCI research
vestigating HCI issues. This is followed by design
findings to design practice. Consider a typical in-
recommendations and research opportunities for
terface design based upon many decisions:
each issue in the framework, and our con-
which functions and objects to include; how they
clusions.
are to be labeled and displayed; whether the in-
terface should use command language, menus,
or icons; and how online help can be provided.
As will be discussed later, each of these deci- Overview of User Interface
sions involves consideration of complicated, and Framework and Theories
sometimes conflicting, human factors. When all
decisions are considered at once, interface Card, et al. (1983) propose the user's recognition-
design becomes overwhelming. Therefore, our action cycle as the basic behavior for understan-
first objective in writing this article is to separate ding the psychology of HCI. This cycle includes
HCI design into major divisions and identify the three stages: the user perceives the computer
most relevant design goals and human factors. presentation and encodes it, searches long and
In each division, design subtasks are analyzed short-term memory to determine a response, and
within the context of current HCI research. The then carries out the response by sending his or
intent of this classification is to assist designers her motor processors in motion. A more elaborate
in relating the research findings to the HCI design seven-stage HCI model is proposed by Norman
process. (1986) (see Figure 1). Norman's model expands
the memory stage to include mental activities,
Early research emphasized the development of such as interpretation and evaluation of system
design guidelines. But, after attempts to both response, formulation of personal goals and in-
write and use guidelines, it was recognized that tentions, and specification of action sequences.
when a design is highly dependent upon task Four cognitive processors are employed in the
context and user behavior, the usefulness of elaborated recognition-action cycle: motor
guidelines diminishes (Gould and Lewis, 1985; movements, perception, cognition, and memory
Moran, 1981). The answer to this problem for a (Olson and Olson, 1990). Except for long-term
particular design is to model the behavior of users memory, these processors have limited capaci-
doing specific tasks. The model provides a basis ty and constrain users' behavior and, thus, HCI
for analyzing why a design works or fails. This design. Most obvious is the need to satisfy users'
leads to the emphasis of understanding cognitive motor and perceptual needs: signals must be
processes employed in HCI; Model Human Pro- perceivable, and responses should be within the
cessor (Card, et al., 1983), SOAR (Laird, et al., range of a user's motor skills. But more impor-
1987), and Task Action Grammars (Payne and tantly, the interface must empower the memory
528 MIS Quarterly/December 1991
Figure 1. Physical and Mental Processes in Operating a Computer

(Adapted from Norman, 1986, and reprinted from Olson and Olson, 1990,
p. 229, by permission of Lawrence Erlbaum Associates)
and cognitive capacity of its users to learn and Overview of the framework
reason easily about the system's behavior. Other-
wise, the user interface will hinder the user's abili- While HCI objectives are clear, it is less obvious
ty to learn all aspects of the system; a bad how the designer should go about developing in-
interface means the user will not use the system terfaces that meet these objectives. Recent
to solve new, difficult problems. research suggests that a system model be
employed as the basis of HCI design (Norman, clear and consistent interfaces. This is the
1986). The system model is a conceptual depic- premise of the interface design framework
tion of the set of objects, permissible operations described in Figure 2. The conceptual aspect of
over the objects, and relationships between ob- the framework concerns design of the system
jects and operations underlying the interface model such that the underlying process the com-
(Jagodzinski, 1983). puter is performing is directly pertinent to the user
in a manner compatible with the user's own
Norman (1986) points out that the selection of a understanding of that process (Fitter, 1979). The
good system model enables the development of physical aspect of the framework involves the
Expectation -l Eva ua ti onli j

AtiSystem
( Model: \
\Ii^^^-- -- -- ^*Task analysis
nterpretat^^r i*Metaphor/Abstrac
Interpretation
rwshX.X....?_Pcm~~;PFV~~?O~
Conceptual Design
Physical Design
Presentation
Language:
*Object representation
*Presentation format
*Spatial layout
*Attention and confirmation
*User assistance
Action Language:
*dialog style
*syntax
*protection mechanism
Figure 2. The HCI Design Framework
design of action and presentation languages, resembles linear algebra with a Fortran-like pro-
which consist of patterns of signs and symbols gramming language; 1-2-3's resembles a paper
enabling the user to communicate to and from spreadsheet and an electronic calculator. The
the system (Bennett, 1983). system model choice results in clear differentia-
tion in the action and presentation languages of
Designing action and presentation languages
these two packages. IFPS's action language re-
based on a coherent system model enables the
quires the user to follow strict syntax rules to
user to easily develop a mental model of the
enter a spreadsheet model. Its presentation is
system through repetitive use. The mental model
that of an accounting report that can only be
is the user's own conceptualization of the system
viewed in a top-down manner. Also, user actions
components, their interrelations, and the process
and system presentations are clearly disjointed
that changes these components (Carroll and
in IFPS; that is, the user first enters the algebraic
Olson, 1988). The mental model provides predic-
formulae, waits for the system to process them,
tive and explanatory power for understanding the
and receives the output when the system is
interaction, enabling the user to reason about
finished.
how to accomplish goals (Halasz and Moran,
1983; Norman, 1986). Hence, the closer the In contrast, 1-2-3's action and presentation
system model is matched to user expectations,
languages are intertwined. 1-2-3 allows the user
the more easily and quickly user learning takes
to enter the spreadsheet by moving to any cell,
place. Developing the system model, therefore,
row, or column in any order to enter data or
requires a study of what the user expectations
specify formulae. Its presentation utilizes the
are. same row-column format used for input; the user
obtains an instant result for each action. The pro-
A system model provides direction for designing
perties of 1-2-3's action and presentation
action and presentation languages that deter-
mine the system's look and feel. When therelanguages
is are more generally accepted than
those
close correspondence between the system model of IFPS, even though both provide similar
capability. Hutchins, et al. (1986) attributes the
and these two languages, the user can manipu-
late all parts of the system with relative ease.success
This of spreadsheet packages like 1-2-3 to
creates an interface of "naive realism" their use of a conceptual model that matches the
user's understanding of spreadsheet tasks.
(diSessa, 1985): one that the user operates
unaware of the computational technicalities
Cognitive
embedded in the system software. But this naive modeling
realism cannot be easily achieved because
As previously mentioned, developing the system
technological restrictions limit the choice of dialog
model requires a study of user expectations. One
style and impose rigid syntax rules and recovery
approach is to create prototypes, which provide
procedures. Hence, in specifying an action
an environment for testing and refining the
language, design tradeoffs must be made be-
system model. This, however, is expensive and
tween satisfying the user's cognitive re-
time-consuming. Alternatively, several cognitive
quirements and satisfying technological
models can be used to analyze and clearly
constraints. The presentation language com-
describe user behavior. This type of theoretical
plements the action language by displaying the
analysis can help designers select the best
results of system execution such that the user
design from several alternatives, resulting in less
can easily evaluate and. interpret the results. It
time needed for HCI design (Lewis, et al., 1990).
also involves design tradeoffs in choosing pro-
per object representations, data formats, spatial
layout, confirmative mechanisms, and user
assistance facilities. GOMS Model
Note that in Figure 2 the system model servesA family of cognitive models based on the GOMS
as the basis for developing action and presenta- model is proposed by Card, et al. (1983) for
tion languages. The importance of this principlepredicting user performance. A GOMS model
is illustrated by the user interfaces of two spread-consists of four cognitive components: (1) goals
sheet packages: IFPS (Execucom, 1979) and and subgoals for the task; (2) operators, in-
1-2-3 (Lotus, 1989). IFPS's system model cluding both overt operators (like key presses)
and internal operators (like memory retrieval); Discussion

(3) methods composed of a series of operators
GOMS, SOAR, and formal grammars collective-
for achieving the goals; and (4) selection rules
ly provide guidance in the design of system
for choosing among competing methods to
models and action and presentation languages.
achieve the same goal. The majority of GOMS
For example, GOMS suggests that system model
research has centered on the study of experts
design should be guided by analysis of user goals
performing well-learned, repetitive tasks. This has
in order to identify methods for achieving these
led to the discovery of parameters, such as times
goals; SOAR demonstrates the importance of
for keystroke entry and the scanning of system
modeling user knowledge of the system model
outputs, useful for predicting skilled-user perfor-
for solving new, difficult problems; TAG indicates
mance (Card, et al., 1983). But other important
how an action language's organization affects
aspects of user behavior cannot be easily mod-
user learning.
eled in GOMS, such as the production of and
recovery from errors (Olson and Olson, 1990) and It should be noted that each of these theories can
the use of sub-optimal goals or methods in per- explain some, but not all, aspects of human
forming routine editing tasks, even when more behavior in HCI. For example, the GOMS model
efficient goals or methods are known (Young, et can explain the task of selecting an option from
al., 1989). a list of choices, but it fails to predict errors a per-
son makes when using a line editor; TAG pro-
vides a reason why errors might occur but cannot
SOAR predict moment-by-moment performance. In ad-
dition, psychological attributes, such as
SOAR (Laird, et al., 1987) is a general cognitive
preference and attitude, and cognitive functions,
architecture of human intelligence. Although it such as mental imagery and cognitive style, are
has not been applied extensively in HCI research,
not considered in these theories (Olson and
SOAR has the potential for answering questionsOlson, 1990). The specificity of each of these
not addressed by GOMS. SOAR is an applica-theories results in areas of uncertainty in HCI
tion of artificial intelligence that models users do-
design, restricting our ability to apply them to
ing both routine and new tasks. In addition to practice.
a A great need for integrating theory and
knowledge base and an engine that performs practice remains in HCI research.
tasks it knows, SOAR has a learning mechanism.
It provides an account of how a user evaluates
system responses and formulates a new goal or
intention. With SOAR, one can estimate how long
System Model Design
it takes a user to recognize an impasse in his orCentral to the entire HCI design question is the
her skill and set up a new goal and action se- design of the system model, a conceptual
quence to overcome that impasse. description of how the system works. This re-
quires an analysis of user tasks so the system
model can be organized to match the user's
Formal Grammars understanding of these tasks (Carroll and
Thomas, 1982; Halasz and Moran, 1982; Moran,
Formal grammars expressed in Backus-Naur
1981). It also requires an analysis of metaphors
form (BNF) can be used to describe the rules of
and abstract models that can adequately portray
an action language. From these, an analyst can
system functionality (Carroll, et al., 1988). The
predict the cognitive effort needed to learn the
result of the latter analysis may also help in selec-
language by examining the volume and con-
ting representations for system objects/functions
sistency of the rules (Reisner, 1981). Task Ac-
and in user training.
tion Grammars (TAG) are similar languages,
which make explicit the knowledge needed for
a user to comprehend the semantics and syntax Analysis of task
of a user interface (Payne, et al., 1986). In addi- The work by Card, et al. (1983) and Norman
tion to identifying the consistency of grammar (1986) indicates that during computer interaction,
rules, TAG can be applied to study how well the the user's mental activities center around goal
task features of the language match user goals. determination and action planning. To ensure
that the system model supports these activities, ports low-level processing by organizing spread-
task analysis should emphasize identifying user sheets into cells, rows, and columns; operations
goals and the methods and objects employed to like "delete" can be applied to any of these data
achieve these goals (Grudin, 1989; Phillips, et al., levels with simple cursor movement and the
1988). same menu action choices.
High-level processing is top-down and is guided

Work Activities and Scenarios by user goals and motives; planning is slow,
serial, and conscious (Newell and Simon, 1972;
Goals, methods, and objects can be discovered
Rasmussen, 1980). A plan of action is a goal
by analyzing users acting out work-related structure that describes how the user decom-
scenarios (Young, et al., 1989). A scenario poses
is a the problem into a sequence of methods
record of a user interacting with some device in
which, when executed, properly handles the work
response to an event, which is carefully con-situation. When facing a complex task, a user
structed so that the user performs a definite ac-
may divide the entire task into many subtasks
tion (like reordering paragraphs of a document
and perform these subtasks separately at dif-
or computing the return on a financial invest-
ferent times (diSessa, 1986). Thus, to support
ment). A carefully constructed set of events higher-level processing, one must ensure that
assures that a comprehensive range of situations
nearly all user goals can be easily achieved
is studied and the results are applicable to brief,
through combinations of operations described in
real-life work situations (Young and Barnard, the system model in either a sequential or
1987). Scenario analysis produces records distributed
of manner. This flexibility can be seen
user actions from which specific user goals,
in Xerox's Star Workstation, where operations for
methods, and objects needed to achieve these
one goal (like creating a document) can be easi-
goals are identified. In addition, records of several
ly suspended to perform operations for another
users completing the same scenario enable the
goal (like creating a spreadsheet) (Bewley, et al.,
designer to compare different approaches to the
1983). Star also allows the user to cut a portion
same work situation and generate a set of of one object (like a spreadsheet) and paste it to
methods and objects for a wide range of users.
another object (a document) to achieve a higher-
level goal of creating a report.
Task analysis results can be documented using

Routine Tasks and Complex Work
GOMS, BNF, TAG, or SOAR. To complete the
Task analysis proceeds by studying cognitive pro-
interface design, details of the methods and the
cesses involved in handling the events. Resear-
operations to be performed on the objects need
chers have observed that users' mental to be specified later during physical design.
processes occur at two levels (Bobrow, 1975).
Low-level processing involves well-learned,
rehearsed procedures for handling routine opera-
Analysis of metaphors and abstract
tions such as data entry or word deletion. High-
models
level processing, which relies upon knowledge
of the system model, is used to generate In plans
designing the system model, it is beneficial to
of action to handle non-routine tasks.
search for metaphors analogical to the system
To support low-level processing, objects need to model. Presenting metaphors to users helps
be organized into logical chunks, and operations them relate the concepts in the system model to
need to match the actions users normally make those already known by a wide set of users. This
with these objects in the real world (Phillips, et enables the user to make inferences regarding
al., 1988). In so doing, learning to associate what system actions are possible and how the
operations with objects is easy; with practice, system model will respond to a given action.
operations can be applied almost automatically,
and even in parallel, because examination of data
content and the meaning of each user action is Metaphors and Composite Metaphors
unnecessary (Shiffrin and Schneider, 1977). For Metaphors can be drawn from tools and systems
example, the spreadsheet system model sup- that are used in the task domain and the
common-sense real world (Carroll, et. al., 1988). tion of interactions among memory, instructions,
For example, many use a typewriter as a input, and output can provide a useful high-level
metaphor for a word processor. Unfortunately, description of a BASIC program's execution.
the analogy between a word processor and a
typewriter breaks down for depicting block inser- Applying Metaphors and Abstract Models
tion and deletion in word processing. For these Metaphors and abstract models are powerful
actions, the word processor works more like a means for conveying the system model to
magnetic tape splicer. Hence, complex systems novices. Mayer (1981) reports that novices who
can be more completely described by a com- lack requisite knowledge are aided by learning
posite of several metaphors, each examined abstract models, which enable them to
closely for its correspondence to the system's ac- understand system concepts during interactions
tual goal-action sequence. Since users general- with the system. Sein and Bostrom (1989) find
ly develop disjointed, fragmented models to that abstract models work best for novices who
explain different kinds of system behavior are able to create and manipulate mental images.
(Waren, 1987), it is easy for them to accom- For other novices, the metaphor is better. Hence,
modate composite metaphors in learning the the choice between metaphor and abstract model
system (Carroll and Thomas, 1982). is dependent upon the user's task knowledge and
Even with composite metaphors, mismatches the ability to conceptually visualize the system
model.
may still occur. Typical computer systems are
more powerful than manual tools and may con-
tain features not embodied in the metaphors, and In conceptual design, candidate metaphors and
vice versa. These mismatches may lead the user abstract models can be identified to provide the
to form misconceptions about how the system designer with building blocks for constructing a
works (Halasz and Moran, 1982). For example, consistent, logical system model based upon the
in word processing, document changes need to user's task model (Waren, 1987). But basing the
be saved or the entire work session is lost; there system model entirely on metaphors may be too
is no such concept applicable to typewriters. Ex- limiting for harnessing the full power of the com-
plicitly pointing out the mismatches to the user puter. The designer's objective should be to pro-
should prevent such misconceptions (Carroll, et perly balance the users' descriptive model of the
al., 1988.) task, the normative model of how the task ought
to be done, and the new opportunities provided
Abstract Models by computer technology.
Abstract models explicitly represent a system

model as a simple, abstract mechanism, which
Iterative system model development
methodologies and tools
the user can mentally "run" to generate expected
system responses (Young, 1981). For example, Task and metaphor analysis must be user-
a hierarchical chart depicting the organization ofcentered and iterative. Initial attempts produce
messages, folders, and files serves as the a crude system model; iterative design and
abstract model of storage for an electronic mail testing rework this crude model into a successful
system, while a file cabinet serves as the system model. For example, questionnaires help
metaphor (Sein and Bostrom, 1989). Like adetermine the basic attributes of the user group
metaphor, the abstract model is not intended to like age, computer training, and education. In-
fully document every detail of a system model; terviews can be used to identify the basic system
rather, both provide a semantic interpretation and
capabilities (Olson and Rueter, 1987). Other
a framework to which the user can attach each
useful approaches include psychological scaling
new system concept (Carroll, et. al., 1988; Mayer,methodologies and simulation and protocol
1981). But unlike a metaphor, there is a one-to- analysis.
one mapping from the attributes of an abstract
model to those of the system model, although not
Psychological Scaling Methodologies
vice versa. Abstract models are particularly useful
for depicting system models that have no real-To identify the grouping of objects/methods, the
designer can solicit user similarity judgments on
world counterparts; for instance, a pictorial depic-
all pairs of objects/operations based upon user Discussion

judgment of frequency of occurrence, temporal
distance, or spatial distance (McDonald and Much research is still needed if we are to
Schvaneveldt, 1988). From this similarity mea- thoroughly understand system model design. Our
surement, clusters of objects/methods can be knowledge of cognitive processes in HCI is still
identified by applying psychological scaling limited, although recent emphases in this area
methodologies, such as hierarchical clustering, indicate an increasing awareness of its
multidimensional scaling, and network structur- significance among researchers and practitioners
ing techniques (e.g., Pathfinder) (McDonald et al., (Olson and Olson, 1990). One important strategy
1988; Olson and Rueter, 1987). These method- is to apply theories like GOMS, TAG, and SOAR
ologies can be applied to organize system docu- to study a broad range of computer tasks for
mentation or menu hierarchy. For example, understanding mental activities involved in solv-
Kellogg and Breen (1987) developed users' views ing routine and novel problems. An attempt at this
of how various elements of documents (footnotes, research has been underway; an Al program in-
captions, etc.) are interrelated; McDonald and cooporating means-ends analysis and multiple
Schvaneveldt (1988) organized UNIX documen- problem spaces has been used to analyze user
tation according to perceived functionality. task knowledge (Young and Whittington, 1990).
This analysis can alert the designer to potential
problems of a proposed interface.
Simulation ana Protocol Analysis Another important strategy is to improve
Requiring users to describe their work re- psychological methods for studying users' prior
quirements in their own language can identify knowledge and cognitive processes. The
useful metaphors and abstract models (Mayer, methods may be applied to investigate how a
1981). Pencil-and-paper simulations of a user forms a mental model of a system and to
proposed interface enable the user to act out evaluate the discrepancies between the user's
typical work scenarios (Gould and Lewis, 1985). mental model and the system model. This pro-
This technique, coupled with think-aloud protocol vides feedback regarding the quality of system
analysis, makes it possible to determine how model design to designers, who can then improve
work is actually done. It is useful for deriving an their design strategies.
initial estimate of the users' set of basic functions
and data objects. In addition, guidance is needed for applying
metaphors to system model design. Whether or
not system models are based upon metaphors,
Another approach is called the Wizard of Oz (Car- users are likely to generate metaphoric com-
roll and Aaronson, 1988). This approach employs parisons on their own (Mack, et al., 1983). What
two linked machines, one for the user and the happens if this comparison creates user confu-
other for the designer. Both the user's display sion because of the discrepancy between the
and the designer's display show a simulated view designer's metaphor choice and the user's own
of the system. To attempt a task, the user enters comparative idea? Strategies are needed for por-
a command, which is routed to the designer's traying metaphors so that the metaphoric com-
screen. The designer simulates the computer by parison is obvious but not distracting. There is
evaluating the user input and sending a response also a need for methodologies for evaluating
to the user's display. This approach has the ad- alternative metaphors. Carroll, et al. (1988)
vantage of putting the user in a work-like situa- hypothesize that the user transforms metaphors
tion well before the final system is fully into a precise understanding of the system model
programmed. Finally, user interface management via a three-stage process: (1) establishing a
systems like GUIDE, Domain/Dialog, and Pro- metaphoric comparison; (2) elaborating aspects
totyper (Hartson and Hix, 1989) or hypermedia of the metaphoric comparison map meaningful-
tools like Hypercard (Halasz, 1988) can be used ly to the system model; and (3) consolidating to
for rapid prototyping to evaluate user needs. They produce a system model from what was learned
are, however, more expensive than the Wizard from each comparison. However, it is unclear
of Oz in terms of manpower and time needed for how this theory can be applied to analyze
creating the prototype. metaphor learnability.
Finally, user confusion may arise when system tasks can vary as much as 9:1 (Egan, 1988). This
concepts have no analogical descriptions, such variance in user performance can be partially at-
as the difference between a line wraparound and tributable to individual differences such as skill
a hard carriage control. How can abstract models level, technical aptitude, age, and cognitive style.
be useful in these situations? Research is need-
The level of user experience and technical skill
ed to provide principles to guide the development
is a dominant factor in selecting an appropriate
of abstract models and strategies for using these
dialog style (Mozeico, 1982). For novices,
models effectively in user training.
computer-guided, constrained-choice interfaces
are better because the time spent on mental ac-
tivities, shown in Figure 1, is reduced. Converse-
Action Language Design ly, with experience comes a clear understanding
The next component of the HCI framework to be of how tasks can be achieved, decreasing the
need for a computer-guided interface and creat-
addressed is the action language design. It in-
volves the creation of a means for the user to ing a preference for a user-initiated language.
easily translate his or her intentions to actions ac- Direct manipulation styles, like Star's iconic desk-
cepted by the system. Because natural language top interface, are easy to learn because they
is not yet a viable option, designers must rely closely reflect the system model, which in turn
upon dialog styles unnatural to novices, relying closely matches the user's task knowledge. They
primarily on keyboards and pointing devices. are easy to use for both novices and experts
Designers must also choose a syntax and vocab-because of simple push-button actions and a con-
ulary for action specifications, and mechanisms tinuous display of the "system states" that guide
for protecting the user from unintentionally user actions (Shneiderman, 1987). Still, direct
destroying completed work. manipulation styles may be slower than conver-
sational styles for experts to use (Hutchins, et al.,
1986).
Dialog style
Novices can become expert through experience.
Many conversation-based dialog styles have This transition is easier if the user possesses
been employed in HCI. In Table 1, these styles technical aptitude, which involves high spatial
are classified according to who inititates the memory and visualization and/or deductive
dialogs and choices available for action specifica- reasoning ability. These abilities help the user
tions (Miller and Thomas, 1977). Recently, direct remember, visualize, and locate objects and
manipulation styles using pointing and graphics generate syntactically correct instructions (Egan
devices have become popular; they differ from 1988).
conversational styles in many aspects (see Table
Cognitive style and age also affect the dialog style
2) (Hutchins, et al., 1986; Shneiderman, 1987). decision. A study by Fowler, et al. (1985) shows
The system model, when designed in accord with that field-independent users, autonomous and
user perception of how tasks are conducted, may self-reliant, prefer a user-initiated command
suggest the dialog style. For example, the "form" structure, while field-dependent users tend to
style is the natural choice for a system involving prefer constrained interfaces. Age is a significant
database inquiries because forms are widely factor in predicting user performance, particularly
used for storing data manually and, as a conse- for interfaces requiring the user to possess a
quence, become the metaphor for that system. technical aptitude (Egan, 1988). The loss in per-
formance due to aging can be countered with a
But choosing a dialog style often requires con-
simplified interface that reduces the necessity of
sidering human factors other than the system
visualizing important displays.
model. The tasks may be complex, suggesting
that no single style is sufficient. For example, ac- Multi-style interfaces can be employed to satisfy
counting application interfaces are often a mix users varying in skill level, cognitive style, and
of forms, menus, and command languages, each age. For example, styles ranging from question-
tailored to specific task requirements. User dif- answer to menu and command language can all
ference also plays an important role. Perfor- be included within the interface; the user can then
mance on relatively low-skill, computer-based choose any style to achieve better performance
Table 1. Taxonomy of Dialog Styles Based on Initiation and Choice
Choice Initiation Free-Response Forced-Choice

User-guided Database language Expert system questions
Command language Input-in-the-context-of-output
Data mnemonics
Text (word) processing
System-guided Question/free answer Question/forced answer
Form filling Command menu selection
Data menu selection
Embedded menu
Accelerated menu
Table 2. Comparison of Conversational and Direct Manipulation Styles
Conversational Style Direct Manipulation Style

Sequential dialog, which requires the user Asynchronus dialog, which
to enter parts of an instruction in a user to enter parts of an instruction in
predetermined order virtually any order
Language of strict syntax to describe the Direct manipulation of ob
user intention
Complete specification of user intention is Incremental specification of user intention

required is allowed
Discrete display of states of sy

executions; this includes errors i
command fails to execute messages are needed
Single-threaded dialogs, which force the Multi-threaded dialogs, which per
user to perform tasks serially user to switch back and forth between
tasks
Command first, object next is typical Object first, command next is typical
Modes are often used to increase keystroke Modeless user operations, which are less
efficiency confusing to the user
and satisfaction (Mozeico, 1982).

theRecently,
action an im-
sequence of enterin
plementation integrating naturalneedslanguage with
to be specified so it can be
direct manipulation (Cohen, nized et al., 1989)
and and
remembered by users.
another combining command language and
direct manipulation (Gerlach and Kuo, 1991)
Vocabulary
show the practicality of this approach.
One way to select vocabulary is f
User interface syntax select keywords based upon the
This approach to vocabulary desig
In interacting with a computer, the user
tuitively is re-
appealing, is shown to
quired to translate his or her goals and intentions
because designers' word choices v
into actions understood by the system.
cantly Hence,
among themselves and ma
in syntax design, designers must users' select words
choices (Carroll, 1985). B
that not only represent system suggests
objects and func-
user testing for obta
tions but also match user expectations.
words. Likewise,
Novices prefer general, frequently used words other qualifier (Bowden, et al., 1989). A single set
that are not representative of system concepts of rules determines how these unique keystroke
(Black and Sebrechts, 1981; Bloom, 1987). Dif- components can be combined to form com-
ferent novices often assign different words to the mands. For example, in a word processing
same concept (Good, et al., 1984; Landauer, et system, commands must obey the rule: first,
al., 1983). As a result, words used by some operation (e.g., DELETE); next, object (e.g., LET-
novices may not help others learn the action TER); and last, direction qualifier (e.g., RIGHT).
language. In an orthogonal language, keystrokes per com-
mand increases in proportion to the size of the
A better alternative is to have expert users select
command set; more time is therefore needed to
terms that are highly representative of system enter commands. But less effort is needed to
concepts; these terms can then be evaluated by
memorize and recall each keystroke's meaning.
novices for learnability (Bloom, 1987). To accom-
This reduction in mental effort and time may
modate both novices' and experts' preferences,
make the memorability-efficiency tradeoff
synonyms should be included as a part of the ac-
beneficial if ease of learning is critical to the user.
tion language (Good, et al., 1984). The alternative
word choices, even if synonyms are not im-
plemented, can be presented to novice users for Action Efficiency
learning the concept of the chosen word (Bloom, Many system implementations concentrate on
1987). minimizing keystrokes to reduce motor activities
through the use of function keys, command ab-
breviations, and recognition of an option's first
Action Consistency letter. But as noted earlier, keystroke efficiency
is also a function of memorizing and recalling the
Consistent keystrokes within and across different
keystrokes. For example, when a function key is
systems lend themselves to easy memorization,
resulting in faster, easier learning. This helps given multiple meanings whose interpretation
users in transferring knowledge of a well-learned
depends upon the context in which it is applied,
a user can be easily confused because of the in-
system to a new system (Polson, 1988; Poison,
creased mental load in recall (Morland, 1983). Of-
et al., 1986). It also reduces user errors and the
time and assistance needed to enter commands fering both whole and abbreviated commands is
one way to increase motor efficiency while reduc-
(Barnard, et al., 1981).
ing the mental load. With these options, the user
Action inconsistency typically occurs in systems can initially enter the whole command and then
employing modes. For example, line editors quickly make use of abbreviated commands
typically have two modes: one for input and the (Landauer, et al., 1983). The importance of reduc-
other for editing. Modes are confusing to novices ing the mental load is further illustrated by Lerch,
because identical keystroke sequences generate et al.'s (1989) study of spreadsheet users perfor-
different results in different modes (Norman, ming financial planning tasks. They found that
1983). However, they are efficient for applications users perform better using relative referencing
in which the number of commands exceeds the of spreadsheet variables (e.g., PREVIOUS
number of keys available. With practice, modes REVENUES) than when using absolute row and
allow experts to use fewer keystrokes for com- column coordiantes. Absolute row and column
mand entry; elimination of modes may penalize coordinates require less keystroke time to enter
the experienced user. Norman recommends that but additional mental overhead. Overall, relative
modes be employed judiciously. We suggest that referencing schemes reduce user errors and
techniques for focusing user attention (discussed allow the user to devote mental capacity to plan-
later) should be used to make modes obvious to ning the task solution.
the user to reduce confusion.
Another way of increasing efficiency is for a
An action language's consistency is affected by system to offer multiple methods for doing the
its orthogonality. In an orthogonal language, each same type of task; the efficiency of each method
basic keystroke component is assigned a unique varies in accordance with the task situation. But
meaning representing a single action parameter, the user may fail to choose the method that re-
which can be an operation, an object, or any quires the least number of keystrokes for a given
task because of the additional mental cost ex-for experts. This issue requires further research
pended in choosing between two methods (Olson in understanding the user's cognitive processes
and Nilsen, 1987). Further investigation mayfor memorization and recall when interacting with
focus on trade-off decisions between using a well- a computer.
rehearsed single general method and learning
and employing several context-specific methods.
Another research issue concerns how to design
an interface or suite of interfaces to satisfy all
users. For example, multi-style interfaces can be
Protection mechanisms created so all styles are equally functional. The
user can then express the same intention in his
The majority of beginners act recklessly; they or her preferred style. To do so, research must
make little effort to read user manuals to acquireaddress questions related to how interfaces can
system knowledge. A survey shows that trial-and- assist users in transferring knowledge from one
dialog style to another. How can one build multi-
error learning is most widely used (Hiltz and Kerr,
1986). A major concern, therefore, is to ensurestyle interfaces so that mastery of one style is in-
that the action language protects the user fromstrumental and perhaps sufficient to facilitate pro-
being penalized for trying the system. gress to another? Can users move from a style
that is system-initiated to one that is user-
One common technique for this is to provide the
intiated? Future research should focus on under-
user with an "undo" function that reverses a
series of actions. Another is to prompt the user
standing cognitive processes for knowledge
transfer, building on the work by Kieras, et al.
to reconsider planned actions that can lead to
(e.g., Kieras and Bovair, 1984; Kieras and Polson,
damaging, irreversible results, such as deleting
a file. 1985).
A third, more interesting approach is "trainingFinally, there is a need for developing principles
wheels," which encourage novices to explore to guide the use of speech and gesture devices.
system features during the initial learning stage
Preliminary studies have shown that users prefer
while protecting them from disaster (Carroll and
these devices (Hauptmann, 1989; Weimer and
Carrithers, 1984). They block invocation of non-
Ganapathy, 1989). Effective incorporation of such
elementary system features and respond withdevices
a in the action language requires further
message stating that the feature is unavailable.
studies to assess their impact on the motor, sen-
The "training wheels" approach effectively sup-sory, perceptual, and cognitive processes of the
ports exploratory learning by reducing the
user.
amount of time users spend recovering from their
errors. But they do not help the learner acquire
system concepts needed for performing tasks not
attempted previously (Catrambone and Carroll, Presentation Language
1987). Research is needed to study what users
Design
learn or do not learn from their mistakes. Another
The last section of the HCI framework concerns
interesting question is the effect of combining the
abstract model and the "training wheels" ap-
presentation language design. An important
proach for providing the user with an interfacedesign objective is for interface displays to guide
for learning the system model. We hypothesize user actions (Bennett, 1983). This objective re-
this combination will result in deeper user quires selecting representations that fit the user's
understanding of system concepts. task knowledge; the format of data produced by
the system must satisfy task needs and prefer-
ences. A display's layout is to be organized so
Discussion that the collective presentation of various outputs
eases user perception and interpretation. Presen-
An important issue of action-language designtations also convey feedback to attract the user's
concerns trade-offs between efficiency and con-attention and confirm user actions. Finally, online
sistency. Keystroke consistency may increaseassistance must be designed to help users learn
learnability for novices but decrease efficiency
system operations and correct their errors.
Object representation et al. (1988) demonstrate that if information is

presented in a format difficult for the user to com-
If the presentation is to adequately reflect the prehend, the user may employ an easier but less
metaphors on which the system model is based, effective decision strategy than one that requires
the designer must choose a display appearance more sophisticated reasoning but leads to a bet-
that assists users in establishing the analogy be- ter result. Lohse (1991) shows that graphs and
tween that display and the metaphors. A familiar tables differ in their cognitive effort. Lohse's
appearance enables the user to recognize and research is interesting because it is based on a
interpret the representation easily. Examples of cognitive model that includes perceptual stores,
this principle are found in the spreadsheet-like short-term memory, algorithms for discrimination
interfaces of 1-2-3 and the electronic desk top of and encoding, and timing parameters. The model
Star. can predict the time needed for a user to unders-
tand a graph. It can be an advisory tool for choos-
Icons can represent much information and be
ing formats to match task needs and has the
easily differentiated (Blattner, et al., 1989). An
potential to answer questions regarding how and
icon can be a concrete picture replicate of a
when graphs and tables can be applied to
familiar object, such as the trash can icon in Star.
faciliate problem solving.
System concepts having no pictoral replicates
can be depicted by abstract icons composed of
geometric shapes and figures. Concrete and
abstract icons may also be combined to create
Spatial layout
hybrid icons, e.g., Ix for deleting a character. User productivity is enhanced when all needed
Unlike concrete icons, abstract and hybrid icons information is readily available. To display as
must be taught to the user. Once learned, much information as possible in a limited area,
however, they are effective on conveying impor- the designer should consider information chunk-
tant system concepts. ing, placement consistency, and the use of win-
dows and 3-D displays.
Presentation formats: table vs.

graph Chunking
Presenting results in graph or table formats to The display, partitioned into well-organized
satisfy both user decision style and task re- chunks that match the user's expectations and
quirements is of great interest to designers of natural perception abilities, provides a basis for
decision support systems. When the task re- the user to select and evaluate actions (Mehlen-
quires a large volume of data, graphs are more bacher, et al., 1989). Chunks can be identified
effective than tables for allowing the user to sum- using the psychological techniques discussed in
marize the data (Jarvenpaa and Dickson, 1988). the system model section. The layout can be
Graphs are also good for tasks (such as interpola- organized following Gestalt principles: the prirn-
tion, trend analysis, and forecasting) that require ciples of proximity and closure suggest enclos-
identification of patterns from large volumes of ing each chunk of objects in a separated area;
data. Conversely, if the task requires pinpointing the principle of similarity suggests using the same
data with precision, tables are better. Tables also font or color for objects of the same chunk. Also,
outperform graphs for simple production schedul- spatial consistency of chunks is important
ing decisions. But for complex decisions, graphs because memorization of location is effortless
are superior (Remus, 1984; 1987). Finally, com- (Mandler, et al., 1977); labels can be used with
bining graph and table formats can result in bet- chunking to improve recognition and recall
ter decisions, albeit with slower performance, (Burns, et al., 1986; Jones and Dumais, 1986).
compared to using either display alone (Powers,
et al., 1984).
Placement Consistency
Our understanding of the cognitive processes in-
volved in handling tables and graphs is still One way proposed to reduce the time in searching
limited. Johnson and Payne (1985) and Johnson, menu items is arranging menus according to fre-
quency of use (Witten, et al., 1984). But this ap- Attention and confirmation
proach may have only a short-term advantage
over a menu with fixed configuration; it may even Video and audio effects are useful in drawing a
cause slower performance because the mental user's attention to important system responses
effort for searching the menu increases with and confirming user actions. Both are important
change and the user becomes disoriented for helping the user judge the status of his or her
(Somberg, 1987; Trevellyan and Browne, 1987). actions.
In the long term, a fixed configuration facilitates
searching better than, or as well as, a dynamic People typically have an orienting reflex to things
menu. The fixed configuration lends itself to that change in their visual periphery. Hence,
memorization, and, therefore, menu selection is video effects such as color, blinking, flashing, and
effortless once it is learned by the user. brightness contrast can stimulate user curiosity
for critical information (Benbasat, et al., 1986;
Morland, 1983). Audio effects can be used to
Windows and 3-D Displays complement video effects or reveal information
difficult to represent with video (Gaver, 1986;
A window is a clearly defined portion of the 1989). In addition, audio feedback can reduce
screen that provides a working area for a par-
space needs and synchronize user input with
ticular task. Windowing has several benefits. Us-
system response (Nakatani, et al., 1986).
ing multiple windows enables the user to
simultaneously perform multiple tasks that may Often there is delay between user actions and
be unrelated. The content of the unfinished task
system presentations. In this situation, confirm-
in a window is preserved so the user can easily atory feedback, such as immediate cursor
continue that task later. Windows also serve as
response and changing shapes and shades of
visible memory caches for integrating informa- icons, is useful (Bewley, et al., 1983; Gould, et
tion from multiple sources or monitoring changes al., 1985). Similarly useful are progress indicators
in separate windows. These benefits collective-
to display the percentage of work completed.
ly enable windowing to support separate but con- Graphic-based progress indicators, like a
current task execution.
percent-done thermometer or a clock, are con-
A drawback of windowing is that operating multi- sidered fun to use (Myers, 1985). Progress in-
ple windows demands higher cognitive pro- dicators also aid in conducting multiple tasks. For
cesses, i.e., memory, perception, and motor example, a user informed that a long time is re-
skills. Overuse of windows can cause informa- quired for printing a document may decide to
tion overload and loss of user control such that spend that time editing another file or retrieving
the user may employ an inefficient search a cup of coffee.
strategy in scanning multiple windows (Hen-
Both visual and auditory cues are shown to
drickson, 1989). Window manipulation is also
shown to be difficult for the user, probably caused
motivate users to explore unknown system
features (Malone, 1984). Incorporating both video
by the complexity in arranging windows (Car-
and audio feedback may have significant impact
roll and Mazur, 1986). Users perform tasks more
on user learning and satisfaction. Auditory icons,
slowly, although more accurately, with windows
(Hendrickson, 1989). Thus, operations for or "earcons," provide intuitive ways to use sound
for presenting information to users (Blattner, et
managing windows should be simplified. The win-
al., 1989; Gaver, 1986; 1989). Like visual icons,
dow design should employ consistent placement
and avoid overcrowded window to ease user auditory icons can be constructed by digitizing
natural sounds with which the user is familiar;
perception and memory load.
abstract auditory icons can also be created by
Also, 3-D displays can be used to accommodate composing a series of sound pitches (Blattner,
and condense a large volume of data (Card, et et al., 1989). For example, in SonicFinder (Gaver,
al., 1991). A 3-D display is divided into many 3-D 1989), a wooden sound is used for opening a file
rooms, each used for a distinct application. The and a metal sound for opening an application,
user can manipulate objects in the 3-D space to while a scraping sould indicates the dragging of
differentiate images, investigate for hidden infor- an object. The research in this area could focus
mation, and zoom in for details. on creating game-like interfaces that are fun to
learn (Carroll and Mazur, 1986) and on assisting

1990). To do so, the designer conducts a GOMS
visually impaired users. analysis of user tasks. The result is then applied
to organize the manual based on possible user
User assistance goals; for each goal, specific "how-to-do- it" in-
formation on methods and operators is then pro-
Three types of information have been shown to
vided. Error avoidance and recovery information
be valuable for providing user assistance (Car-
can be included to improve user performance.
roll and Aaronson 1988; Kieras and Bovair, 1984).
One is "how-to-do-it" information that defines
specific action steps for operating the system.
Another is "what-it-is-for" information that Query-in-Depth
elaborates on the purpose of each step; this helpsQuery-in-depth is a technique designed to pro-
users associate steps with individual goals. Thirdvide multi-level assistance to help users at
is "how-it-works" information that explains thevarious levels of expertise learn the system
system model; this is useful for advanced trouble-(Gaines, 1981; Houghton, 1984). Its low-level help
shooting and creative use of the system. All three includes brief "how-to-do-it" and "what-it-is-for"
types of information can be used in writing onlineinformation that instructs users' immediate ac-
error messages and user instructions. tions. If not satisfied, the user can request more
advanced "how-it-works" information for trouble-
shooting.
Error Correction
When novices make errors and are uncertain Discussion

about what to do next, they often look for instruc-
tions from the system message (Good, et al.,
In the past 10 years, engineers have created
sophisticated video and audio technologies for
1984). Thus, error messages should pinpoint cor-
rective, "how-to-do-it" information and state computer input and output. New technologies,
like Virtual Reality and Speech I/O, will likely be
"what-it-is-for" (Carroll and Aaronson, 1988). In
addition, immediate feedback on user errors integrated into normal presentations. To effec-
tively apply them, we need to better understand
facilitates learning better than delayed feedback
how they affect the user in performing work.
because a user can easily associate the correct
Studies have shown that while auditory memory
action with the exact point of error (Catrambone
has less storage capacity than visual memory,
and Carroll, 1987). The style of error messages
it retains signals more than twice as long as visual
is also important: they should reflect users'
memory (Cowan, 1984). These differences in at-
words, avoid negative tones, and clearly identify
tention and memory phenomena must be ex-
the portion of the action in error (Shneiderman,
amined within the context of human-computer
1987).
interaction. What is the impact on user cognitive
process given that only limited capacity is
available for motion and perception? How should
Online Manuals
the various devices be integrated? What are the
When users know the task they wish to perform, costs and benefits in terms of hardware, software,
brief "guided exploration cards" (Catramboneuser training, and actual user performance? Pro-
and Carroll, 1987) help users perform better than viding guidance in designing video and audio in-
long manuals. Specific "how-to-do-it" informa- terfaces is challenging but critical in HCI research
tion can be included for novices to do complete in the near future.
tasks quickly in the begkRning (Carroll and Aaron-
Windowing offers many advantages in action and
son, 1988; Catrambone, 1990). In addition, in-
presentation language design that have yet been
structions describing general rules of the system
explored. For example, one way to implement
model encourage novices to infer unstated details
multi-style interfaces is to allow each style to be
of the interface, resulting in better user learning
operated in a separate window. Or, to adapt to
of the system (Black, et al., 1989).
a user's pattern of menu usage, a window for the
The GOMS model described earlier can be used most recently used menu options, another for the
to create online manuals (Gong and Elkerton, most frequently used options, and a third for the
regular menu options can be used in combina- research is needed to expand these theories so
tion. Windows are ideal for user assistance: er- they can be useful in addressing a wide range
ror messages, online manual, or confirmatory of interface design issues based upon user and
feedback can be located in windows separated task considerations.
from work dialogs. Complex tasks can also be
Great challenges remain ahead in interface
supported by allowing subtasks in separate win- research. We should not limit ourselves to the
dows or 3-D rooms. Again, research is needed
study of problems concerning only existing
to study how windows and 3-D rooms can be ef-
technologies. We should explore new, creative
fectively applied for these various purposes. The
uses of advanced technologies to know what,
central issue is to understand how they can im-
when, and how to apply them effectively. We can
pact the user's cognitive processes, as discussed
save substantial research effort by ceasing to em-
in the work by Card, et al. (1991).
phasize problems inherent in poorly developed
Finally, there is a need for research in online ad- technologies unless they illuminate cognitive pro-
vising. Research so far has shown that online ad- cesses that will be important to interfaces of the
vising, even that provided by an expert using the future (Wixon, et al., 1990).
Wizard-of-Oz technique, is of limited use for the
We need to broaden research concerning how
novice user (Carroll and Aaronson, 1988). The
difficult issues to be addressed are what infor-
people organize, store, and retrieve concepts
(Carroll and Campbell, 1986; Newell and Card,
mation should be given and when, what ideas
1985; 1986). Theories of exemplar memory, pro-
should be left to user inference, and how to use
totype memory, episodic memory, and seman-
motivational feedback to make learning en-
tic memory are probably applicable to HCI
joyable. Studies could also explore the use of
research. We also need to investigate
video and audio feedback in assisting the user.
psychological attributes (such as attitude and
preference), work-related factors (such as fatigue
Conclusion and organizational culture), and certain physical
limitations (such as hearing and vision impair-
Interfaces are complex, cybernetic-like systems
ment). We must study how user interfaces should
that can be built quickly but are difficult to build
cope with the limitations imposed by varying user
well. Their complexities necessitate the decom- characteristics. More importantly, we must focus
position of the entire user-interface design pro-on what aspects of user characteristics are im-
blem into small, manageable subproblems, along portant, how they are related to each stage of HCI
with a reexamination of their interrelationships in-
design, and when during the design stage they
to a whole. The framework presented in this arti-
must be considered. This focus ensures the ap-
cle serves this purpose; it organizes research
plicability to research findings to design.
findings into three major divisions: system model,
action language, and presentation language. This
Finally, we must interrelate the research findings
if we are to develop comprehensive theories for
article reviews current HCI research findings and
illuminates their practical implications. The aim
the design, implementation, and testing of func-
of this work is to enable HCI design practicetional,
to usable, and learnable interfaces. In this
become more systematic and less intuitive than
pursuit, the role of the designer in documenting
it is today. his or her design rationales is especially impor-
Throughout the literature two major philosophies tant. A design rationale is a record of design after-
of interface design and research can be iden- natives and explanation of why some specific
tified. One is that interface design is often driven choice is made. To further our understanding of
by technological advancement; research is con- HCI, design rationales should be a co-product of
ducted to address problems that occur after a the design process (Maclean, et al., 1989). Com-
design is implemented. This generated the paring and contrasting design rationales of
mouse, voice, windows, and graphics. The other various systems enables us to capture the range
is that we still know little about the psychological of constraints affecting the HCI design and gain
insights into why a choice works or does not work.
make-up of the user. The work on the psychology
of HCI by Card, et al. (1983) and Norman (1986) Some excellent exploratory work has been done
provide a solid theoretic beginning; much in this area. For example, Wixon, et al. (1990) pro-
pose collection of usability data in the context of Design of Xerox's 8010 STAR Workstation,"
user tasks to identify both general principles and Proceedings of CHI'83 Human Factors in
detailed guidelines for HCI design. Carroll and Computing Systems, Boston, MA, 1983, pp.
Kellogg (1989) and Carroll (1990) emphasize the 72-77.
identification of psychological claims embodied Black, J.B., Bechtold, J.S., Mitrain, M, and Car-
in an interface and the application of artifacts as roll, J.M. "On-line Tutorials: What Kind of In-
bases for assessing appropriateness of these ference Leads to the Most Effective
claims. In conclusion, data regarding user tasks, Learning?" Proceedings of CHI'89 Hum
user achievement and problems, and changes Factors in Computing Systems, Austin, T
in the overall environment should be collected on 1989, pp. 81-83.
a continuous basis. Assumptions about the Black, J.B. and Sebrechts, M.M. "Facilitatin
psychology of the user performing the task and Human-Computer Communication," Appl
limitations of technology must be explicitly stated. Psycholinguistics (2), 1981, pp. 149-177.
The collection of design rationales can then be Blattner, M.M., Sumikawa, D.A., and Greenbe
used to develop practical guidelines and prin- R.M. "Earcons and Icons: Their Structure and
ciples, which should be repeatedly evaluated to Common Design Principles," Human-
develop theories governing HCI design. Computer Interaction (4:1), 1989, pp. 11-44.
Bloom, C.P. "Procedures for Obtaining and
Testing User-Selected Terminologies,"
Acknowledgements Human-Computer Interaction (3:2),
We are indebted to the anonymous reviewers for 1987-1988, pp. 155-177.
their considerable effort in reviewing this article.
Bobrow, D.G. "Dimensions of Representations,"
We are particularly thankful to the associate in Representation and Understanding, D.G.
editor, Judith Olson, for her insights into the field
Bobrow and A. Collins (eds.), Academic Press,
of HCI. Their many recommendations contributed
New York, NY, 1975, pp. 1-34.
Bowden, E.M., Douglas, S.A., and Stanford, C.A.
significantly to this article's development.
"Testing the Principle of Orthogonality in
Language Design," Human Computer Interac-
tion (4:2), 1989, pp. 95-120.
References
Burns, M.J., Warren, D.L., and Rudisill, M. "For-
matting Space-Related Displays to Optimize
Barnard, P.J. "Command Names," in Handbook Expert and Nonexpert User Performance,"
of Human-Computer Interaction, M. Helender, Proceedings of CHI'86 Human Factors in
(ed.), Elsevier Science Publishers, Amster- Computing Systems, Boston, MA, 1986, pp.
274-280.
dam, 1988, pp. 181-199.
Card, S.K., Moran, T.P., and Newell, A. The
Barnard, P.J., Hammond, N. V., Morton, J., Long,
J.B., and Clark, I.A. "Consistency and Com- Psychology of Human-Computer Interaction,
patibility in Human/Computer Dialogue," In- Lawrence Erlbaum Associates, Hillsdale, NJ,
ternational Journal of Man-Machine Studies 1983.
(15), 1981, pp. 87-134. Card, S.K., Robertson, G.G., and Mackinlay, J.D.
Benbasat, I., Dexter, A.S., and Todd, P. "An Ex- "The Information Visualizer: An Information
perimental Program Investigating Color- Workspace," Proceedings of CHI'91 Human
Enhanced and Graphical Information Presen- Factors in Computing Systems, New Orleans,
tation: An Integration of the Findings," Com- LA, 1991, pp. 181-188.
munications of the ACM (29:11), December Carroll, J.M. "The Adventure of Getting to Know
1986, pp. 1094-1105. a Computer," IEEE Computer (15:11), Nov.
Bennett, J. "Analysis and Design of the User In- 1982, pp. 49-58.
terface for Decision Support Systems," in Carroll, J.M. What's in a Name, Freeman, New
Building Decision Support Systems, J. Ben- York, NY, 1985.
nett (ed.), Addison-Wesley, Reading, MA, Carroll, J.M. "Infinite Detail and Emulation in an
1983, pp. 41-64. Ontologically Minimized HCI," Proceedings of
Bewley, W.L., Roberts, T.L., Schroit, D., and CHI'90 Human Factors in Computing
Verplank, W.L. "Human Factors Testing in the Systems, Seattle, WA, 1990, pp. 321-327.
Carroll, J.M and Aaronson, A.P. "Learning by Do- tegrated Computational Environment,"
ing with Simulated Intelligent Help," Com- Human-Computer Interaction (1:2), 1985, pp.
munications of the ACM (31:9), September 1-47.
1988, pp. 1064-1079. diSessa, A.A. "Models of Computation," in User
Carroll, J.M. and Carrithers, C. "Training Wheels Centered System Design, D.A. Norman and
in a User Interface," Communications of the S.W. Draper (eds.), Lawrence Erlbaum
ACM (27:8), August 1984, pp. 800-806. Associates, Hillside, NJ, 1986, pp. 201-218.
Carroll, J.M. and Campbell, R.L. "Softening Up Egan, D.E. "Individual Differences in Human-
Hard Science: Reply to Newell and Card" Computer Interaction," in Cognitive Science
Human Computer Interaction (2:3), 1986, and its Application for Human-Computer In-
pp. 227-249. teraction, H. Helendar (ed.), Elsevier Science
Carroll, J.M. and Kellogg, W.A. "Artifact as Publishers B.V., Hillsdale, NJ, 1988, pp.
Theory-Nexus: Hermeneutics Meets Theory- 543-568.
Based Design," Proceedings of CHI'89, Execucom Systems Corporation. Cases and

Human Factors in Computing Systems, Models Using IFPS, Execucom, Austin, TX,
Austin, TX, 1989, pp. 7-14. 1979.
Carroll, J.M., Mack R.L., and Kellogg, W.A. "In- Fitter, M. "Towards More Natural Interactive
terface Metaphors and User Interface Systems," International Journal on Man-
Design," in Handbook of Human-Computer In- Machine Studies (11:3), 1979, pp. 339-350.
teraction, M. Helander, (ed.), Elsevier Science Fowler, C.J.H., Macaulay, L.A., and Fowler, J.F.
Publishers, Amsterdam, 1988, pp. 67-86. "The Relationship Between Cognitive Style
Carroll, J.M. and Mazur, S.A. "Lisa Learning," and Dialogue Style: An Explorative Study," in
IEEE Computer (19:11), November 1986, pp. People and Computers: Designing the Inter-
35-49. face, P. Johnson and S. Cook (eds.), Cam-
Carroll, J.M. and Olson, J.R. "Mental Models in bridge University Press, New York, NY, 1985,
Human-Computer Interaction," in Handbook pp. 186-198.
of Human-Computer Interaction, M. Helander Gaines, B. "The Technology of Interaction-
(ed.), Elsevier Science Publishers, Amster- Dialogue Programming Rules," International
dam, 1988, pp. 45-65. Journal of Man-Machine Studies (14:1), 1981,
Carroll, J.M. and Thomas, J.C. "Metaphor and pp. 133-150.
the Cognitive Representation of Computing Gaver, W. "Auditory Icons: Using Sound in Com-
Systems," IEEE Transactions on Systems, puter Interfaces," Human-Computer Interac-
Man, and Cybernetics (12:2), 1982, pp. tion (2:2), 1986, pp. 167-177.
107-116. Gaver, W.W. "The SonicFinder: An Interface that
Catrambone, R. "Specific Versus General Pro- Uses Auditory Icons," Human-Computer In-
cedures in Instructions," Human Computer In- teraction (4:1), 1989, pp. 67-94.
teraction (5:1), 1990, pp. 49-93. Gerlach, J.H. and Kuo, F.Y. "Formal Develop-
Catrambone, R. and Carroll, J.M. "Learning a ment of Hybrid User-Computer Interfaces with
Word Processing System with Training Advanced Forms of User Assistance," Jour-
Wheels and Guided Exploration," Pro- nal of Systems and Software (16:3), November
ceedings of CHI + GI 1987 Human Factors 1991, pp. 169-184.
in Computing Systems, Toronto, Ontario,
1987, pp. 169-174.
Gong, R. and Elkerton, J. "Designing Minimal
Documentation Using a GOMS Model: A
Cohen, P.R., Dalrymple, M., Moran, D.B.,
Usability Evaluation of an Engineering Ap-
Pereira, F.C.N., Sullivan, J.W., Gargan, R.A.,
proach," Proceedings of CHI'90 Human Fac-
Jr., Schlossberg, J.L., and Tyler, S.W.
tors in Computing Systems, Seattle, WA,
"Synergistic Use of Direct Manipulation and
1990, pp. 99-106.
Natural Language," Proceedings of CHI'89
Human Factors in Computing Systems, Good, M.D., Whiteside, J.A., Wixon, D.R., and
Austin, TX, 1989, pp. 227-234. Jones, S.J. "Building a User-Derived Inter-
Cowan, N. "On Short and Long Auditory Stores," face," Communications of the ACM (27:10),
Psychological Bulletin (96), 1984, pp. 341-470. October 1984, pp. 1032-1043.
diSessa, A.A. "A Principled Design for an In- Gould, J.D., Lewis, C., and Barnes, V. "Cursor
Movement During Text Editing," ACM Tran- Centered System Design, D.A. Norman and
sactions on Office Information Systems (3:1), S.W. Draper (eds.), Lawrence Erlbaum
January 1985, pp. 22-34. Associates, Hillsdale, NJ, 1986, pp. 87-124.
Gould, J.D. and Lewis, C. "Designing for Usabili- Jagodzinski, A.P. "A Theoretical Basis for the
ty: Key Principles and What Designers Think," Representation of On-Line Computer Systems
Communications of the ACM (28:3), March to Naive Users," International Journal of Man-
1985, pp. 300-311. Machine Studies (18), 1983, pp. 215-252.
Grudin, J. "The Case Against User Interface Jarvenpaa, S.L. and Dickson, G.W. "Graphics
Consistency," Communications of the ACM and Managerial Decision Making: Research-
(32:10), October 1989, pp. 1164-1173. Based Guidelines," Communications of the
Grudin, J. "The Computer Reaches Out: The ACM (31:6), June 1988, pp. 764-774.
Historical Continuity of Interface Design," Pro- Johnson, E.J. and Payne, J.W. "Effort and Ac-
ceedings of CHI'90 Human Factors in Com- curacy in Choice," Management Science
puting Systems, Seattle, WA, 1990, pp. (31:4), April 1985, pp. 395-414.
261-268. Johnson, E.J., Payne, J.W., and Bettman, J.R.
Halasz, F.G. "Reflections on Notecards: Seven "Information Displays and Preference Rever-
Issues for the Next Generation of Hypermedia sals," Organizational Behavior and Human
Systems," Communications of the ACM Decision Processes (42), 1988, pp. 1-21.
(31:7), July 1988, pp. 836-852. Jones, W.P. and Dumais, S.T. "The Spatial
Halasz, F.G. and Moran, T.P. "Analogy Con- Metaphor for User Interfaces: Experimental
sidered Harmful," Proceedings of the Con- Tests of Reference by Location versus
ference on Human Factors in Computing Names," ACM Transactions on Office Infor-
Systems, Gaithersburg, MA, 1982, pp. mation Systems (4:1), January 1986, pp.
383-386. 42-63.
Halasz, F.G. and Moran, T.P. "Mental Models
Kellogg, W.A. and Breen, T.J. "Evaluating User
and Problem Solving in Using a Calculator,"
and System Models: Applying Scaling Tech-
Proceedings of CH1'83 Human Factors in
niques to Problems in Human-Computer In-
Computing Systems, Austin, TX, 1983, pp.
212-216.
teraction," Proceedings of CHI + GI 1987
Human Factors in Computing Systems,
Hartson, H.R. and Hix, D. "Human-Computer In-
Toronto, Ontario, 1987, pp. 303-308.
terface Development: Concepts and Systems
Kieras, D.E. and Bovair, S. "The Role of Mental
for Its Management," Computing Surveys
Knowledge in Learning to Operate a Device,"
(21:1), March 1989, pp. 5-92.
Cognitive Science (8), 1984, pp. 191-219.
Hauptmann, A.G. "Speech and Gestures for
Kieras, D.E. and Poison, P.G. "An Approach to
Graphic Image Manipulation," Proceedings of
the Formal Analysis of User Complexity," In-
CHI'89 Human Factors in Computing ternational Journal of Man-Machine Studies
Systems, Austin, TX, 1989, pp. 241-245.
(22), 1985, pp. 365-394.
Hendrickson, J.J. "Performance, Preference,
and Visual Scan Patterns on a Menu-Based Laird, J.E., Newell, A., and Rosenbloom, P.S.
System: Implications for Interface Design," "SOAR: An Architecture for General In-
Proceedings of CHI'89 Human Factors in telligence," Artificial Intelligence (33), 1987,
Computing Systems, Austin, TX, 1989, pp. pp. 1- 64.
217-222. Landauer, T.K., Galotti, K.M., and Hartwell, S.
Hiltz, S.R. and Kerr, E.B. "Learning Modes and "Natural Command Names and Initial Learn-
Subsequent Use of Computer-Mediated Com- ing: A Study of Text-Editing Terms," Com-
munication Systems," Proceedings of CHI'86 munications of the ACM (26:7), July 1983, pp.
Human Factors in Computing Systems, 495-503.
Boston, MA, 1986, pp. 149-155. Lerch, F.J., Mantei, M.M., and Olson, J.R.
Houghton, R.C. "Online Help Systems: A Con- "Skilled Financial Planning: The Cost of
spectus," Communications of the ACM (27:2), Translating Ideas into Action," Proceedings
February 1984, pp. 126-133. of CHI'89 Human Factors in Computing
Hutchins, E.L., Hollan, J.D. and Norman, D.A. Systems, Austin, TX, 1989, pp. 121-126.
"Direct Manipulation Interfaces," in User Lewis, C., Poison, P., Wharton, C., and Rieman,
J. "Testing a Walkthrough Methodology for (9), 1977, pp. 509-536.

Theory-Based Design of Walk-Up and Use In- Moran, T. "An Applied Psychology of the User,"
terfaces," Proceedings of CHI'90 Human Fac- Computing Surveys (13:1), March 1981, pp.
tors in Computing Systems, Seattle, WA, 1-12.
1990, pp. 235-242. Morland, D.V. "Human Factors Guidelines for
Lewis, M.W. and Anderson, J.R. "Discrimination Terminal Interface Design," Communications
of Operator in Problem Solving: Learning from of the ACM (26:7), July 1983, pp. 100-104.
Examples," Cognitive Psychology (17), 1985, Mozeico H. "A Human/Computer Interfaces to
pp. 26-65. Accommodate User Learning Stages," Com-
Lohse, J. "A Cognitive Model for the Perception munications of the ACM (25:2), February 1982,
and Understanding of Graphs," Proceedings pp. 100-104.
of CHI'91 Human Factors in Computing Myers, B.A. "The Importance of Percent-Done
Systems, New Orleans, LA, 1991, pp. Indicators for Computer- Human Interfaces,"
137-144. Proceedings of CHI'85 Human Factors in
Lotus Development Corporation. Lotus 1-2-3, Computing Systems, San Francisco, CA,
Lotus Development Corporation, Cambridge, 1985, pp. 11-17.
MA, 1989. Nakatani, L.H., Egan, D.E., Ruedisueli, L.W.,
Mack, R.L., Lewis, C.H., and Carroll, J.M. Hawley, P.M., and Lewart, D.K. "TNT: A Talk-
"Learning to Use Word Processors: Problems ing Tutor 'N' Trainer for Teaching the Use of
and Prospects," ACM Transactions on Office Interactive Computer Systems," Proceedings
Information Systems (1:3), July 1983, pp. of CHI'86 Human Factors in Computing
254-271. Systems, Boston, MA, 1986, pp. 29-34.
Maclean, A., Young, R.M., and Moran, T.P. Newell, A. and Card, S. "The Prospects of
"Design Rationale: The Argument Behind the Psychological Science in Human-Computer
Artifact," Proceedings of CHI'89 Human Fac- Interaction," Human Computer Interaction
tors in Computing Systems, Austin, TX, 1989, (1:3), 1985, pp. 209-242.
pp. 247-252. Newell, A. and Card, S. "Straightening Out
Malone, T.W. "Heuristics for Designing En- Softening Up: Response to Carroll and Camp-
joyable User Interfaces: Lessons from Com- bell," Human Computer Interaction (2:3),
puter Games," in Human Factors in 1986, pp. 251-267.
Computing Systems, J.C. Thomas and M.
Newell, A. and Simon, H.A. Human Problem
Schneider (eds.), Ablex, Norwood, NJ, 1984, Solving, Prentice-Hall, Englewood Cliffs, NJ,
pp. 1-12. 1972.
Mandler, J.M., Seegmiller, D., and Day, J. "On Nickerson, R.S. "Why Interactive Computer
the Encoding of Spatial Information," Memory Systems Are Sometimes Not Used by the Peo-
Cognition (5), 1977, pp. 10-16. ple Who Might Benefit from Them," Interna-
Mayer, R.E. "The Psychology of How Novices tional Journal of Man-Machine Studies (4),
Learn Computer Programming," Computing 1981, pp. 469-483.
Surveys (13:1), March 1981, pp. 121-141. Norman, D.A. "Design Rules Based on Analysis
McDonald, J.E. and Schvaneveldt, R.W. "The of Human Error," Communications of the
Application of User Knowledge to Interface ACM (26:4), April 1983, pp. 254-258.
Design," in Cognitive Science and Its Applica- Norman, D.A. "Cognitive Engineering," in User
tion for Human-Computer Interaction, R. Centered System Design, D.A. Norman and
Guinden (ed.), Lawrence Erlbaum Associates, S.W. Draper (eds.), Lawrence Erlbaum
Hillsdale, NJ, 1988, pp. 289-338. Associates, Hillsdale, NJ, 1986, pp. 31-61.
Mehlenbacher, B., Duffy, T.M., and Palmer J. Olson, J.R. and Nilsen, E. "Analysis of the Cogni-
"Finding Information on a Menu: Linking tion Involved in Spreadsheet Software Interac-
Menu Organization to the User's Goals," tion," Human-Computer Interaction (3:4),
Human-Computer Interaction (4:3), 1989, pp. 1987, pp. 309-349.
231-251. Olson, J.R. and Olson, G.M. "The Growth of
Miller, L.A. and Thomas, J.C., Jr. "Behavior Cognitive Modeling in Human Computer In-
Issues in the Use of Interactive Systems," In- teraction Since GOMS," Human Computer In-
ternational Journal of Man Machine Studies teraction (5:2-3), 1990, pp. 221-266.
Olson, J.R. and Rueter, H.H. "Extracting Exper- Learning, Automatic Attending, and a General
tise from Experts: Methods for Knowledge Ac- Theory," Psychological Review (84:2), March
quistion," Journal of Expert Systems (4:3), 1977, pp. 127-190.
1987, pp. 152-168. Shneiderman, B. Designing the User Interface,
Payne, S.J. and Green, T.R.G. "Task-Action Addison- Wesley, Reading, MA, 1987.
Grammars: A Model of the Mental Represen- Somberg, B.L. "A Comparison of Rule-Based
tation of Task Languages," Human-Computer and Potentially Constant Arrangements of
Interaction (2:2), 1986, pp.93-134. Computer Menu Items," Proceedings of CHI
Phillips, M.D., Howard, B.S., Ammerman, H.L., + GI 1987 Human Factors in Computing
and Fligg, C.M., Jr. "A Task Analytic Ap- Systems, Toronto, Ontario, 1987, pp. 255-260.
proach to Dialogue Design," in Handbook of Trevellyan, R. and Browne, D.P. "A Self-
Human-Computer Interaction, M. Helander Regulating Adaptive System," Proceedings of
(ed.), Elsevier Science Publishers, Amster- CHI + GI 1987 Human Factors in Computing
dam, 1988, pp. 835-857. Systems, Toronto, Ontario, 1987, pp. 103-107.
Poison, P., Muncher, E., and Englebeck, G. "A Waren, Y. "Mental Models in Learning Com-
Test of a Common Elements Theory of puterized Tasks," in Psychological Issues of
Transfer," Proceedings of CHI'86 Human Human Computer Interaction in the Work
Factors in Computing Systems, Boston, MA, Place, M. Frese, E. Ulich, and W. Dzida (eds.),
1986, pp. 78-83. Elsevier Science Publishers, Amsterdam,
Poison, P. "The Consequences of Consistent 1987, pp. 275-294.
and Inconsistent Interfaces," in Cognitive Weimer, D. and Ganapathy, S.K. "A Synthetic
Science and Its Application for Human- Com- Visual Environment with Hand Gesturing and
puter Interaction, R. Guinden (ed.), Lawrence Voice Input," Proceedings of CHI'89 Human
Erlbaum Associates, Hillsdale, NJ, 1988, pp. Factors in Computing Systems, Austin, TX,
59-107. 1989, pp. 235-240.
Powers, M., Lashley, C., Sanchez, D., and Witten, I.H., Cleary, J., and Greenberg, S. "On
Shneiderman, B. "An Experimental Com- Frequency-Based Menu-Splitting Algorithms,"
parison of Tabular and Graphical Data International Journal of Man-Machine Studies
Presentation," International Journal of Man- (21), 1984, pp. 135-148.
Machine Studies (20), 1984, pp. 545-566. Wixon, D., Holtzblatt, K., and Knox, S. "Contex-
Rasmussen, J. "The Human as a System Com- tual Design: An Emergent View of System
ponent," in Human Interaction with Com- Design," Proceedings of CHI'90 Human Fac-
puters, H.T. Smith and T.R.G. Green (eds.), tors in Computing Systems, Seattle, WA,
Academic Press, London, 1980, pp. 67-96. 1990, pp. 329-336.
Reisner, P. "Using a Formal Grammar in Human Young, R.M. "The Machine Inside the Machine:
Factors Design of an Interactive Graphics Users' Models of Pocket Calculators," Inter-
System," IEEE Transactions on Software national Journal on Man-Machine Studies
Engineering (7:2), March 1981, pp. 1409-1411. (15), 1981, pp. 51-85.
Remus, W. "An Experimental Investigation of the Young, R.M. and Barnard, P.J. "The Use of
Impact of Graphical and Tabular Data Presen- Scenarios in Human-Computer Interaction
tations of Decision Making," Management Research: Turbo-Charging the Tortoise of
Science (30:5), May 1984, pp. 533-542. Cumulative Science," Proceedings of CHI +
Remus, W. "A Study of Graphical and Tabular GI 1987 Human Factors in Computing
Displays and Their Integration with En- Systems, Toronto, Ontario, 1987, pp. 291-296.
vironmental Complexity," Management Young, R.M., Barnard, P., Simon, T., and Whit-
Science (33:9), September 1987, pp. tington, J. "How Would Your Favourite User
1200-1204. Model Cope with These Scenarios?" SIGCHI
Sein, M.K. and Bostrom, R.P. "Individual Dif- Bulletin (20:4), April 1989, pp. 51-55.
ferences and Conceptual Models in Training Young, R.M. and Whittington, J. "Using a
Novice Users," Human-Computer Interaction Knowledge Analysis to Predict Conceputal
(4:3), 1989, pp. 197-229. Errors in Text-Editor Usage," Proceedings of
Shiffrin, R.M. and Schneider, W. "Controlled and CHI '90 Human Factors in Computing
Automatic Information Processing: Perceptual Systems, Seattle, WA, 1990, pp. 91-97.
About the Authors
James H. Gerlach is associate professor of in- Feng-Yang Kuo is assistant professor of infor-
formation systems at the University of Colorado mation systems in the Graduate School of
at Denver. In addition to human-computer in- Business, University of Colorado at Denver. He
teraction, his research interests include software received his Ph.D. in management information
engineering and EDP auditing. His work has ap- systems from the Univerity of Arizona. His
peared in ACM Transactions on Information research interests include human-computer in-
Systems, IEEE Computer, Decision Support teraction, database management, office automa-
Systems, Journal of Systems and Software, The tion, and decision support systems. Dr. Kuo's
Accounting Review, and Auditing. Dr. Gerlach work has appeared in MIS Quarterly, Com-
received an M.S. in computer science and a munications of the ACM, Information Manage-
Ph.D. in management, both from Purdue ment, and Decision Support Systems.
University.

Understanding Human-Computer Interaction For Information Systems Design

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Understanding Human-Computer Interaction For Information Systems Design

Uploaded by

Copyright:

Available Formats

Understanding Human-Computer Interaction for Information Systems Design

Author(s): James H. Gerlach and Feng-Yang Kuo

Management Information Systems Research Center, University of Minnesota is

design is divided into three major divisions:

Computer Interaction tion language. The system model is a concep-

MIS Quarterly/December 1991 527

528 MIS Quarterly/December 1991

Figure 1. Physical and Mental Processes in Operating a Computer

MIS Quarterly/December 1991 529

Expectation -l Eva ua ti onli j

Figure 2. The HCI Design Framework

530 MIS Quarterly/December 1991

MIS Quarterly/December 1991 531

and internal operators (like memory retrieval); Discussion

532 MIS Quarterly/December 1991

High-level processing is top-down and is guided

Task analysis results can be documented using

MIS Quarterly/December 1991 533

Abstract models explicitly represent a system

534 MIS Quarterly/December 1991

all pairs of objects/operations based upon user Discussion

MIS Quarterly/December 1991 535

536 MIS Quarterly/December 1991

Table 1. Taxonomy of Dialog Styles Based on Initiation and Choice

Choice Initiation Free-Response Forced-Choice

Table 2. Comparison of Conversational and Direct Manipulation Styles

Conversational Style Direct Manipulation Style

Complete specification of user intention is Incremental specification of user intention

Discrete display of states of sy

and satisfaction (Mozeico, 1982).

MIS Quarterly/December 1991 537

538 MIS Quarterly/December 1991

MIS Quarterly/December 1991 539

Object representation et al. (1988) demonstrate that if information is

Presentation formats: table vs.

540 MIS Quarterly/December 1991

MIS Quarterly/December 1991 541

learn (Carroll and Mazur, 1986) and on assisting

When novices make errors and are uncertain Discussion

542 MIS Quarterly/December 1991

MIS Quarterly/December 1991 543

544 MIS Quarterly/December 1991

Based Design," Proceedings of CHI'89, Execucom Systems Corporation. Cases and

MIS Quarterly/December 1991 545

546 MIS Quarterly/December 1991

J. "Testing a Walkthrough Methodology for (9), 1977, pp. 509-536.

MIS Quarterly/December 1991 547

548 MIS Quarterly/December 1991

About the Authors

MIS Quarterly/December 1991 549

You might also like