Professional Documents
Culture Documents
net/publication/227998407
CITATIONS READS
28 27,339
1 author:
Andrew Dillon
University of Texas at Austin
181 PUBLICATIONS 5,314 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Andrew Dillon on 26 February 2019.
CONTENTS
Introduction Cognitive theories and models in HCI
Human±computer interaction Developing user-centered design methods
Cognitive design guidelines Conclusion
The design of computer interfaces that are usable involves physical input and visual perception of
and easily learned by humans is a non-trivial prob- output, any basic interactive device must incorpor-
lem for software developers. As information tech- ate an input device and a screen. However, to de-
nologies mediate many of the activities we now termine the appropriate input the user must have
perform routinely, the process of human±computer
some representation of a goal or intention of an
interaction is of fundamental importance.
outcome to attain, necessitating the employment
of memory, both short-term (to handle current
INTRODUCTION status information) and long-term (to enable the
planning and interpretation of the interactive
Since much human±computer interaction (HCI) is sequence).
cognitive in nature, involving perception, represen- Information undergoes transformation at each
tation, problem-solving, navigation, query formu- stage of progression from a perceived stimulus
lation and language processing, the theories and (e.g. a visual change in the interface) to a compre-
methods of cognitive science are directly relevant hended cue (e.g. a recognized sign), leading to an
to it. Thus an applied cognitive science for software active response (e.g. pushing a mouse button). All
design has emerged. of this happens repeatedly and rapidly as the
Traditional cognitive science approaches to HCI human interacts with the system. This process can
and user interface design model the user as made be simply envisioned using a variant of Neisser's
up of three basic components: the psychomotor, perceptual cycle model, whereby users engage in
perceptual, and cognitive subsystems. Recent treat- an ongoing cycle of information exchange involv-
ments of HCI have extended this model to include ing exploration of a changing information environ-
the social system as an essential component of the ment (see Figure 1).
user and have placed greater emphasis on group For example, an interface normally provides a
dynamics and social context in examining what response to user input which signals to the user
users do with technology. Any full treatment of that his or her input has caused an action. Through
user psychology must embrace all these com- direct manipulation, users can select and move
ponents, though cognitive issues dominate most objects on screen, open and close windows with a
research in HCI. mouse click, or jump from place to place within
documents via hyperlinks. In each case, the user
must initiate an action and the interface must com-
HUMAN±COMPUTER INTERACTION
municate its change of state through appropriate
The success of any computer application is depend- feedback to the user. Where feedback is vague, too
ent on it providing appropriate facilities for the rapid or nonexistent, the user is likely to be con-
task at hand in a manner that enables users to fused. Ideally feedback communicates to the user
exploit them effectively. Whereas the provision of that the system status has altered in the intended
facilities is an issue of functionality, the user inter- manner and that the user is closer to his or her goal
face is the means by which the functionality can as a result.
be used, and here we are directly concerned with We can usefully understand HCI as a cognitive
usability. process by considering the human as possessing
For interaction to proceed, the human user must general knowledge structures (e.g. mental models
input a signal to the computer and perceive changes or schemata) which organize the user's task-
in the interface. Since most current interaction oriented information exploration and use. This
454 User Interface Design
concepts used in HCI, metaphor has proved one of only where a process is being explained, although
the most durable and accepted. many designers deliberately exploit the natural
human perceptual tendency to attend to movement
Learning by Doing by using animation to capture attention, particu-
larly for advertisements in commercial internet
The most successful systems are those that enable sites.
a user to get something done as soon as possible.
Users tend to be very resistant to reading any
accompanying documentation and often want to
Using Images and Icons
get on with real tasks immediately rather than Screen `real estate' is a limited commodity, so de-
follow any training guides. Hence, error-free per- signers seek means of conveying concepts and
formance is not considered a real goal. Instead, actions through the medium of signs, images, and
cognitive scientists emphasize the importance of symbols. Another reason for iconic interfaces is
clear and informative feedback, and the ability to their independence of language and their pre-
undo actions, to support the user through the sumed ability to cross cultural boundaries.
learning process. Semiotic approaches to design have been in-
Having gained some knowledge by using one voked to help designers create appropriately com-
part of a system, users will expect to be able to prehensible icons, but the results have been mixed.
apply this throughout the system. Particular atten- Current interfaces make extensive use of graphic
tion should be paid to the consistent use of terms, capabilities and iconic representations but couple
colors, and highlighting techniques, and the pos- these with pop-up text labels that explain their
itioning of task-related zones on the screen, so as to meaning to users who find the representations
support generalization by the user. Consistency difficult to decipher.
between systems can also be important to maintain
± for example, between the old and the new ver-
sions. The benefits of a new system can easily be
COGNITIVE THEORIES AND MODELS
obscured if users feel that their existing knowledge IN HCI
is redundant and they must learn the new system It is not yet possible to talk of a complete theory
from scratch. of human±computer interaction, given the many
activities, processes, and tasks that computers
Minimizing Attentional and Cognitive support. However, to overcome the piecemeal ap-
Load proach that results from repeated empirical tests
of evolving interface features, attempts have been
Some theoretical insights into cognitive architec- made to produce stronger theoretical models to
ture emphasize the memory and attentional con- guide interface designers. This approach has
straints of humans. These lessons have been worked best where it has been constrained to spe-
learned by the HCI community who argue that cific or localized interactive phenomena rather than
interaction sequences should be designed to min- the full range of user responses to information
imize the load on short-term memory (e.g. not technology.
asking a user to choose from an excessive number
of menu items, or requiring the user to remember
Interaction as Serial Information
numbers or characters from one screen to another).
Processing
Since recognition memory is superior to absolute
recall, the use of menus is now the norm in design, Cognitive scientists have derived many findings
as opposed to the command-line interfaces of the about human information processing, and this
1980s, which required users to memorize control knowledge has been distilled in the area of HCI
arguments. into a form of engineering model of the user that
Another related contribution of cognitive science can be exploited by designers. Generally referred
to user interface design has been in the area of task to as the `model human processor', this cognitive
sequencing. User interface designers are encour- model enables interface designers to predict the
aged to minimize the number of steps for which time a user will take to complete a task sequence
information must be retained by the user. Instead, given an analysis of the cognitive, perceptual, and
designers are encouraged to provide all necessary psychomotor components that are applied at each
information in the interface for the user to exploit step. For example, to determine how long it would
as needed. The use of animation is recommended take a user to complete the task sequence involving
User Interface Design 457
saving a file to hard disk, consider the data in experimental findings, each new if±then rule pro-
Table 1, derived from laboratory studies of duction will take a typical user about 25 seconds
humans. to learn. Armed with such knowledge, designers
To apply such a model, the designer would first could estimate, for example, the costs involved in
list the basic steps a user must take with an inter- changing procedures or violating consistency of
face. We can imagine a proposed design that re- interaction with new designs. This is obviously a
quires the user to locate the mouse (Th), move the gross estimate, but for many proceduralized tasks
mouse to a menu (Tp), select the `save' command the data indicate the underlying regularity of
(Tp Tk), allow the system to respond with a human performance.
prompt (Tr), input the filename (Tk (number of
letters)), and then hit a save button (Tk). The de-
Sociocognitive Analyses of HCI:
signer could quickly use the estimates from Table
1 to calculate how long a user would take to per-
Activity Theory and Acceptance Models
form this sequence, and use these data to evaluate An alternative application of cognitive theory has
the proposed design. emerged as HCI researchers have become inter-
The exact values of these estimates can be de- ested in user acceptance of computers and the
bated, but the principle of the model human exploitation of technology by groups of users.
processor is constant; i.e. decompose the task into Such research draws less on the traditional base
its constituent actions and calculate the time in- of laboratory findings within cognitive science
volved in the serial processing of these actions. and more on its social and anthropological trad-
Multiple applications of this method have con- itions. These theories may be called `sociocognitive'
firmed its value in estimating expert or error-free theories.
task completion times for repetitive, nondiscretion- Activity theory aims to bring a closer reading of
ary tasks. cultural forces to bear on our analyses of inter-
However, this model has its limitations. We action. Users are seen as situated within a context
cannot use it to estimate how long users will that exerts strong forces on their actions. Further-
spend on tasks that are not highly practiced, or more, such users are dynamic, changing as their
that require decision-making, planning, or learn- experience and application of technology changes.
ing. Similarly, where tasks involve parallel process- Taking an activity-theoretic approach to HCI, it is
ing, it is easy to overestimate times by assuming important to extend analyses of interface usability
simple serial processing of the task actions. How- to cover the contexts in which the technology is
ever, as an applied model of cognition for a limited used (or rejected).
range of routine and well-practiced tasks, such a Typical activity-theoretic approaches examine
technique is clearly useful. HCI in terms of the praxis, or situated context ±
There have been several extensions of this ap- e.g. a banking organization, a teaching scenario, or
proach, most notably to cover learning. Based on a medical process ± in which the various levels of
a production system analysis (describing the be- interaction can take place, from automatic individ-
havior a user must learn to complete a task in ual operations to collective ventures or activities
terms of a series of `if±then' rules of interactive that define the group's purpose. The analysis and
sequences; e.g. `if file is ªnewº then select menu design of any technology needs to be grounded in
option ªnew fileº '), cognitive complexity theory such a broader perspective to ensure it is appropri-
enables calculation of the estimated time it would ate and usable by the intended user community.
take a user to learn a new procedure. According to One can see activity theory as extending traditional
cognitive approaches rather than replacing them.
Other socially-oriented approaches to HCI that
consider cognition include the general class of ac-
Table 1. Time estimates for completion of basic inter-
ceptance theories that seek to predict whether a
active tasks by a human operator
user, given a choice, will utilize a technology.
Label Action Time estimate (seconds) Such models emphasize the perceived value that
users place on the new technology, and measure
Tk Enter a keystroke 0.23
Th Move hand to mouse 0.36
the relationship between such ratings and subse-
Tp Point mouse 1.5 quent behavior in context. For example, it is now
Tm Retrieve from memory 1.2 known that if users perceive a new tool as having
Tr Computer to respond 1.2 direct usefulness for them in their work, they will
be more likely to choose it, and may tolerate a
458 User Interface Design
certain difficulty of use for the sake of the power it a need for better expert-based evaluation methods
affords them. Such perceptions by users seem to be to overcome the rather poor validity of most
formed very quickly, often within minutes of inter- such methods. (Testers employing these methods
acting for the first time. This means that early tend to overestimate the number of problems
impressions due to aesthetics, implementation users actually experience; that is, they label as
style, and related factors are particularly important. `problems' many aspects of interfaces that users
Theoretical developments in HCI have not kept subsequently perceive as acceptable.) Similarly, re-
pace with developments in technology, partly be- searchers have tried to package formal methods
cause of the speed of technological change, but also into tools that can be used effectively by non-
because of the difficulty of translating cognitive specialists to predict usability. The aim of this ap-
science into rich theoretical models that predict proach is to develop software tools that designers
human behavior in multiple contexts. Neverthe- could use to estimate learning effort or time to
less, although some dismiss the theoretical ap- perform a task, without having to know the details
proach as too limited for practical application, of how such an estimate is derived. This would be
most HCI professionals are of the view that long- analogous to the use engineers can make of the
term progress is possible only with increased effort principles of physics. However, few such tools
at deriving and applying cognitive science theories have yet made the transition from research labora-
to the problems of user interface design. tory to design practice.