Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (1)
The Human
Overview
Human can be viewed as an informationprocessing system, for example, Card, Moran andNewell's Model Human Processor.
A simple model:
� information received and responses givenvia input{output channels
� information stored in memory
� information processed and applied invarious ways
Capabilities of humans in these areas areimportant to design, as are individual di�erences.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (2)
Input{Output channels
Vision
Two stages in vision
� physical reception of stimulus
� processing and interpretation of stimulus
The physical apparatus: the eye
� mechanism for receiving light andtransforming it into electrical energy
� light reects from objects; their images arefocused upside-down on retina
� retina contains rods for low light vision andcones for colour vision
� ganglion cells detect pattern and movement
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (3)
Interpreting the signal
Size and depth
� visual angle indicates how much of �eld ofview object occupies (relates to size anddistance from eye)
� visual acuity is ability to perceive �ne detail(limited)
� familiar objects perceived as constant sizein spite of changes in visual angle | law ofsize constancy
� cues like overlapping help perception of sizeand depth
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (4)
Interpreting the signal (cont)
Brightness
� subjective reaction to levels of light
� a�ected by luminance of object
� measured by just noticeable di�erence
� visual acuity increases with luminance asdoes icker
Colour
� made up of hue, intensity, saturation
� cones sensitive to colour wavelengths
� blue acuity is lowest
� 8% males and 1% females colour blind
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (5)
Interpreting the signal (cont)
The visual system compensates for movementand changes in luminance.
Context is used to resolve ambiguity.
Optical illusions sometimes occur due to overcompensation.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (6)
Figure 1: The Ponzo illusion
Figure 2: The Muller Lyer illusion
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (7)
Reading
Several stages:
� visual pattern perceived
� decoded using internal representation oflanguage
� interpreted using knowledge of syntax,semantics, pragmatics
Reading involves saccades and �xations.Perception occurs during latter.
Word shape is important to recognition.
Negative contrast improves reading fromcomputer screen.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (8)
Hearing
Provides information about environment:distances, directions, objects etc.
Physical apparatus:
� outer ear | protects inner and ampli�essound
� middle ear | transmits sound waves asvibrations to inner ear
� inner ear | chemical transmitters arereleased and cause impulses in auditorynerve
Sound
� pitch | sound frequency
� loudness | amplitude
� timbre | type or quality
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (9)
Hearing (cont)
Humans can hear frequencies from 20Hz to15kHz | less accurate distinguishing highfrequencies than low.
Auditory system �lters sounds | can attend tosounds over background noise. For example, thecocktail party phenomenon.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (10)
Touch
Provides important feedback about environment.
May be key sense for someone who is visuallyimpaired.
Stimulus received via receptors in the skin:
� thermoreceptors | heat and cold
� nociceptors | pain
� mechanoreceptors | pressure (someinstant, some continuous)
Some areas more sensitive than others e.g.�ngers.
Kinethesis | awareness of body positiona�ecting comfort and performance.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (11)
Movement
Time taken to respond to stimulus: reaction time+ movement time
Movement time | dependent on age, �tness etc.
Reaction time | dependent on stimulus type:
� visual | 200ms
� auditory | 150 ms
� pain | 700ms
Increasing reaction time decreases accuracy in theunskilled operator but not in the skilled operator.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (12)
Movement (cont)
Fitts' Law describes the time taken to hit ascreen target:
Mt = a+ b log2(D=S + 1)
where a and b are empirically determinedconstants, Mt is movement time, D is Distanceand S is Size.
Targets in general should be large as possible andthe distances as small as possible.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (13)
Memory
There are three types of memory function.
Iconic EchoicHaptic
Sensory memories
or
Short-term memory
Working memory
Long-term memoryAttention Rehearsal
Sensory memory
Bu�ers for stimuli
� iconic | visual stimuli
� echoic | aural stimuli
� haptic | touch stimuli
Constantly overwritten.
Information passes from sensory to STM byattention.
Selection of stimuli governed by level of arousal.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (14)
Short-term memory (STM)
Scratch-pad for temporary recall
� rapid access | 70ms
� rapid decay | 200ms
� limited capacity | 7 + =� 2 digits orchunks of information
Recency e�ect | recall of most recently seenthings better than recall of earlier items.
Some evidence for several elements of STM |articulatory channel, visual channel etc. |interference on di�erent channel does not impairrecall.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (15)
Long-term memory (LTM)
Repository for all our knowledge
� slow access | 1/10 second
� slow decay, if any
� huge or unlimited capacity
Two types
� episodic | serial memory of events
� semantic | structured memory of facts,concepts, skills
Information in semantic LTM derived fromepisodic LTM.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (16)
Long-term memory (cont.)
Semantic memory structure
� provides access to information
� represents relationships between bits ofinformation
� supports inference
Model: semantic network
� inheritance | child nodes inherit propertiesof parent nodes
� relationships between bits of informationexplicit
� supports inference through inheritance
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (17)
Long-term memory (cont.)
is a DOG
is a
colour: [brown/white, black/white, merle]
instance
is a
HOUND
is a
BEAGLE
instance
SNOOPY
friend of
CHARLIE BROWN
colour:[brown,black/white]
size:small
tracks
has tail
has four legs
SHEEPDOG
works sheepbarks
ANIMAL
breathes
movesis a
COLLIEsize: medium
LASSIE
film character
colour:brown/white
SHADOW
colour: brown/whitebook character
instance
cartoon/book character
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (18)
Long-term memory (cont.)
Other models of LTM
Frames:
Information organized in datastructure. Slots in structure areinstantiated with particular values fora given instance of data.
DOG
Fixedlegs: 4
Default
diet: carnivoroussound: bark
Variablesize:colour:
COLLIE
Fixedbreed of: DOGtype: sheepdog
Default
Variablecolour:
size: 65 cm
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (19)
Long-term memory (cont.)
Scripts:
Model of stereotypical informationrequired to interpret situation orlanguage. Script also has elementswhich can be instantiated withparticular values.
dog needs medicinedog needs operation
Tracks:
payingexamination waiting in roomarriving at receptionScenes:
vet examinesRoles:diagnosestreats
owner brings dog inpaystakes dog out
Props: examination tablemedicineinstruments
Result: dog betterowner poorervet richer
dog illvet openowner has money
Entry conditions:
Script for a visit to the vet
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (20)
Long-term memory (cont.)
Production rules:
Representation of proceduralknowledge. Condition{action rules |if condition is matched, rule �res.
LTM processes
Storage of information
� information moves from STM to LTM byrehearsal
� amount retained proportional to rehearsaltime: total time hypothesis
� optimized by spreading learning over time:distribution of practice e�ect
� structure, meaning and familiarity makeinformation easier to remember
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (21)
LTM processes (cont.)
Forgetting
� decay { information is lost gradually butvery slowly
� interference | new information replacesold: retroactive interference
� old may interfere with new: proactiveinhibition | so may not forget at all
� memory is selective and a�ected by emotion| can `choose' to forget
Information retrieval
� recall | information reproduced frommemory. Can be assisted by cues, e.g.categories, imagery
� recognition | information gives knowledgethat it has been seen before. Less complexthan recall | information is cue.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (22)
Thinking: reasoning and problem solving
Reasoning
Deductive: derive logically necessary conclusionfrom given premises. E.g.
If it is Friday then she will go to workIt is FridayTherefore she will go to work.
Logical conclusion not necessarily true:
If it is raining then the ground is dryIt is rainingTherefore the ground is dry
Human deduction poor when truth and validityclash.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (23)
Reasoning (cont.)
Inductive: generalize from cases seen to casesunseen. E.g. all elephants we have seen havetrunks therefore all elephants have trunks.
Unreliable: can only prove false not true.
However, humans are not good at using negativeevidence. E.g. Wason's cards.
74 E K
Abductive: reasoning from event to cause. E.g.Sam drives fast when drunk. If see Sam drivingfast, assume drunk.
Unreliable: can lead to false explanations.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (24)
Problem solving
Process of �nding solution to unfamiliar taskusing knowledge.
Several theories.
Gestalt
� problem solving both productive andreproductive
� productive problem solving draws oninsight and restructuring of problem
� attractive but not enough evidence toexplain `insight' etc.
� move away from behaviouralism and led toinformation processing theories
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (25)
Problem solving (cont.)
Problem space theory
� problem space comprises problem states
� problem solving involves generating statesusing legal operators
� heuristics may be employed to selectoperators e.g. means-ends analysis
� operates within human informationprocessing system e.g. STM limits etc.
� largely applied to problem solving in wellde�ned areas e.g. puzzles rather thanknowledge intensive areas
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (26)
Problem solving (cont.)
Analogy
� novel problems are solved by usingknowledge from a similar domain in newdomain | analogical mapping
� analogical mapping may be di�cult ifdomains are semantically di�erent
Skill acquisition
Skilled activity characterized by
� chunking | lot of information is chunkedto optimize STM
� conceptual rather than super�cial groupingof problems | information is structuredmore e�ectively
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (27)
Skill acquisition (cont.)
Model of skill acquisition: ACT*
3 levels of skill
� general purpose rules to interpret factsabout problem | knowledge intensive
� speci�c task rules are learned | rely onknown procedures
� rules are �ne-tuned | skilled behaviour
Mechanisms for moving between these
� proceduralization | level 1 to level 2
� generalization | level 2 to level 3
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (28)
Skill acquisition { proceduralization
Level 1:
IF cook[type, ingredients, time]THEN
cook for: time
cook[casserole, [chicken,carrots,potatoes],2 hours]
cook[casserole, [beef, dumpling, carrots],2 hours]
cook[cake, [our, sugar,butter, egg], 45 mins]
Level 2:
IF type is casseroleAND ingredients are [chicken,carrots,potatoes]THEN
cook for: 2 hours
IF type is cakeAND ingredients are [our,sugar,butter,eggs]THEN
cook for: 45 mins
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (29)
Skill acquisition { generalization
Level 2:
IF type is casseroleAND ingredients are [chicken,carrots,potatoes]THEN
cook for: 2 hours
IF type is casseroleAND ingredients are [beef,dumplings,carrots]THEN
cook for: 2 hours
Level 3:
IF type is casseroleAND ingredients are ANYTHINGTHEN
cook for: 2 hours
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (30)
Errors and mental models
Types of error
� slips | change to aspect of skilledbehaviour can cause slip
� incorrect understanding | humans createmental models to explain behaviour. Ifwrong (di�erent from actual system) errorscan occur.
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (31)
Individual di�erences
� long term | sex, physical and intellectualabilities
� short term | e�ect of stress or fatigue
� changing | age
Ask: will design decision exclude section of userpopulation?
Human{Computer Interaction, Prentice Hall
A. Dix, J. Finlay, G. Abowd and R. Beale c1993The HumanChapter 1 (32)
Cognitive Psychology and
Interactive System Design
Some direct applications. E.g. blue acuity is poorso blue should not be used for important detail.
However, application generally requires
� understanding of context in psychology
� understanding of particular experimentalconditions
A lot of knowledge has been distilled in
� guidelines | see Chapters 4 and 5
� cognitive models | see Chapter 6
� experimental and analytic evaluationtechniques | see Chapter 11