1 CMT 3210: Understanding the human element in HCI Lecture 4 Models of human-Computer Interaction Mental Models Elke Duncker.

1

CMT 3210: Understanding the human element in HCI

Lecture 4Models of human-Computer InteractionMental Models

Elke Duncker

2

Models of Human-Computer Interaction

Normans 7 stage modelHuman Processor model

3

Evaluation and execution

Approximate model of how people carry out tasks

Emphasises carrying out actions and getting feedback

The World

Goal

Evaluation: Comparing

what happened

with what we wanted

Execution:Carrying out actions on the world

4

The World

Perception

Cyclic model of human action

Seven stage process Can start anywhere

Goal

Intentionformation

Action specification

Execution

Evaluation

Interpretation

5

Gulfs of execution and evaluation

Gulf of execution: How does the user translate intentions into action?

Gulf of evaluation: How does the user understand the effects of actions

and does s/he tell when her/his goals are satisfied?

Goals World

Execution

Evaluation

6

Bridging the gulfs

Execution Mappings:

are actions designed so the user make the connection between the effects they intend to achieve and the actions provided by the system?

Evaluation Feedback:

is information about the system state provided in a way that allows users to determine whether goals are satisfied?

7

Using the model as a design aid

Tell what actions are possible

Determine mapping from intention to

action

Perform the action

Tell if system in desired state

Determine mapping from state to interpretation

Determine the function of the device

Tell what state the system is in

How easy is it for the user to:

8

Principles for good design

Visibility of the state of device and possibilities for action?

Good conceptual model consistent presentation, coherent system image?

Good mappings relation between actions and effects?

Feedback about effects of actions?

9

A model of human information processing

Model Human ProcessorDetailed model of how computer users process

information Intended to help designers do predictionsConsists of specialised concurrent processing

units and memories characterised by speed, decay time, capacity,

encoding

10

Model Human Processor

Working memory

Long-term memory

Auditory memory

Visual memory

Stimulus

Perception

Cognitive processo

r

Motor processo

r

11

Principles of the MHP

Perception takes time - depends on stimulus intensity

Processing takes time - depends on task demands, information load, practice

Different encoding used at different stagesUsers act rationally (most of the time):

Goals + Task + Operators + Inputs + Knowledge +

Processor limitsBehaviour

12

Applications in HCI

The GOMS and KLM analysis techniques Identifies kinds of processing and memory Identifies possible concurrency

13

Comparing the models

MHP almost exclusively about what goes on inside the head Focus on human performance - especially time

Norman's model more about human-machine interaction without saying how processes are implemented or

what exists in the head Distinguish between "in the head" and "in the world"

14

Knowledge in the head and in the world

Knowledge needed to translate intentions into plans and actions may be remembered by the user (in the head)

e.g. unix commands

embedded in the devices used (in the world) e.g. direct manipulation interfaces

15

Summary HCI models

Norman's 7 stage model of human action Identifies processes involved in action Doesn't specify how they take place Useful as a tool for thinking and analysis Consideration of "in the head" / "in the world"

Model Human Processor Detailed account of human information processing Specifies how internal processes work Used for making predictions (e.g. GOMS)

Framework for topics that will come later

16

Mental Models

A person’s image about an artefactThe user’s image about the machineMaking sense of the world from inside

17

Mental Models

An internal representation of a system that can be interrogated and manipulated.

    Mental models are concrete.     Mental models can be run.     Mental models are constructed from

experience.Mental models are generally incomplete and

inaccurate, but serve a purpose

18

Example Camera

What are the components and how do they fit together?

How does it work? What causes what? How do you use it? How do you use your understanding when

something goes wrong?

19

Example: London Underground Map

• what is it good for? • what isn’t it much use for?

20

The Contents of a Model

Kieras (1982) categorised the kinds of knowledge that people have about a device as follows. Label or name of the device Function or purpose (what goals can be

accomplished) Controls and indicators Inputs, outputs and connections Power sources and requirements

21

Contents of a Model (cont.)

External layout and appearance Internal layout and appearance External behaviour (input-output function) How to operate the device to accomplish goals Procedures for troubleshooting and maintenance Internal structure and mechanisms (how it works)

22

Mental Models

A runnable mental model = how-it-works knowledge + knowledge how-to-use-the-how-it-works-knowledge !

How-it-works knowledge may be at various levels of detail.

Strategic knowledge includes various strategies — e.g. inference, prediction, diagnosis. These are transferable skills.

23

Origin of mental models

a) declarative and procedural modelsb) feedback, hypothesis-testing (but beware

mode errors)c) manuals and help systemsd) analogy and metaphorse) designer’s conceptual model:

24

How designers influence the user’s mental model

Designer’s conceptual

model

Interface design

User’s mental model

25

Note

This view puts the designer at the centre It is the designer’s job is to enable the user to

assimilate their view by getting the device to project an appropriate image of itself.

This is necessary for device-only constructs (where the challenge is to make these as easy to learn and make sense of as possible).

26

Domains and devices

'Mental models' are about understanding the device [and its representation of the domain].

User may be: manipulating domain

through device, or investigating domain

through device

27

Further Reading

For the execution-evaluation model: Donald Norman "The Design of Everyday Things”

(also appeared as "The Psychology of Everyday Things")

For the Model Human Processor: Stuart Card, Thomas Moran and Alan Newell "The

Psychology of Human Computer Interaction"