Lecture 6

IE 486 Work Analysis & Design II
Instructor: Vincent Duffy, Ph.D.
Associate Professor of IE
Lecture 6 – Decision Making & Uncertainty
Thurs. Feb. 1, 2007
An introduction to human decision making
First review end of cognition (lecture 5) and
relationship to decision making (lecture 6)
Re-introduce QOTD as part of lecture today
Then briefly discuss plan for tomorrow in lab
6. Long term memory

Design implications


The user is unlikely to develop the same level of
detail (recall) in the use of a product compared to
the designer
Therefore, when possible, it is a good idea to:

Encourage frequent use of information –

Take advantage of frequency and recency


Standardize and use memory aids


then over time, the task may become more automatic
Eg. Give list of instructions for sending fax
Carefully design information to be remembered

Make it meaningful and avoid the use of technical
jargon when possible
7. Attention and mental resources

If we devote cognitive resources to one activity,
others are likely to suffer

Eg. Study of the use of cell phones and driving
suggests that accidents are 5x more likely


Consider multiple resources

Eg. One can not read a book and watch tv at the
same time.


The rate is roughly equivalent to driving drunk
However, one can listen to the spoken version of the
book while watching tv.
Visual and auditory processing requires separate
resources.
8. Conclusions

Some general design implications

Consider: issues related to




Make the input mode ‘dissimilar’ when possible


Divided attention,
controlled vs. automatic processing &
multiple resources
eg. Take advantage of different pools of resources.
Automation (or more automatic processing of information)
allows better time-sharing of mental resources.

Convey Priority - Let the user know the importance
of each task to better allocate (mental) resources.
Overview of human decision making
1. Overview of human decision making
2. An example: anesthesiology team in hospital
3. What is the problem with heuristics for
decision making?
4. Naturalistic decision making
5. Skill rule and knowledge based task
performance
6. Improving human decision making
IE 486 - Lecture 6 - QOTD
QOTD 1. What are ‘heuristics’?
QOTD 2. How can we model task
performance considering the cognitive
aspects of tasks?
QOTD 3. What are some ways to improve
human decision making through human
factors engineering design?
1. Overview of human decision making
According to Wickens (ch.7)
Decisions are made either
– intuitively - quick and relatively automatic or
– analytically - slow, deliberate and controlled
1. Overview of human decision making
QOTD 1. What are ‘heuristics’?
– Simplifications in decision making
– do not always guarantee best solution due to biases or
misperceptions
– eg. ‘satisficing’ (Simon, a psychologist 1957)
– suggests a decision maker generates alternatives until an
‘acceptable’ (not necessarily optimal) solution is found
it is believed that many people will judge that ‘going beyond this to
identify something better has ‘too little advantage to make it worth
the effort’.
Why?
– People have limited cognitive capacity and limited time
1. Overview of human decision making
How do economists believe we make decisions?
– Rational decision making - it is expected that the
decision maker will find the ‘optimal’ solution
– …based on our concept of what is useful (utility) and
willingness to accept risk
2. An example: anesthesiology team in
hospital
5 medical procedures are to be performed
urgently on 5 different patients in two different
buildings
– there are only 3 anesthesiologists plus the ‘one in
charge’ who is supposed to be ‘available’ in case of
incoming unexpected ‘emergency’
– classical or ‘normative’ decision making theory
suggests there are different alternatives with different
likelihood of outcome …and each has an expected
‘utility’ (good/bad payoff)
2. anesthesiology team in hospital
4 possible alternatives
– the ‘one in charge’ begins a procedure and no
emergency occurs
– or the ‘one in charge’ begins a procedure and an
emergency occurs
– or she doesn’t begin a procedure and an emergency
comes in
– or she doesn’t begin a procedure and an emergency
doesn’t come in
– see p. 160 for expected outcomes based on likelihood
and utility/payoffs
3. What is the problem with heuristics for
decision making?
The decisions are subject to biases
– 1. A limited number of hypotheses is generated
– 2. ‘memory’ research suggests people will recall what
was most ‘frequently’ or most ‘recently’ considered most readily available.
– 3. Certain cues may lead to a conclusion - then not
enough is then done to eliminate other possibilities
– 4. Overconfidence - people tend to believe they are
right more often than they really are
– 5. Cognitive fixation - people tend to ignore cues that
are contrary to their original belief
4. Naturalistic decision making
Decisions in ‘the field’
– these ideas are considered outside the ‘experimental’
world
It is suggested by some that this is more useful
that ‘experimental studies done in labs’.
However, Wickens suggests that these are
complementary to experimental evidence
– For ill structured problems
– under time constraints and time stress
– high risk and multiple people involved
5. Skill, rule and knowledge based task
performance
QOTD 2
Q. How can we model task performance considering the cognitive
aspects of tasks?
Rasmussen suggests 3 levels of ‘cognitive’ control and that people
operate at one of the levels depending on the nature of the task and
their ‘experience’
– skill based - reacting to perceptual elements in an
automatic, subconscious level
– rule based - rely on if-then associations between cues
and actions
typical of those with familiarity but not extensive experience
– knowledge based- when the situation is novel and
there is no rule or previous experience to draw on
6. Improving human decision making
QOTD 3
Q. What are some ways to improve human decision
making through human factors engineering design.
Eg. Parachutist has a chute that fails to open properly
– after trying to untangle the cords, they deploy the reserve chute
too late - at 200 ft.
To reduce chance of an accident
– a redesign could have helped recognition of the trouble and
awareness of the critical ‘time’ issue
different colors for cloth and cords could have helped
diagnosis and
an altitude sensor (with auditory alarm) could have given
awareness about time running out
6. Improving human decision making
in complex environments can be helped by training
– using computer support
– to teach pattern recognition
– break the process into different cognitive steps
‘People overrely on rapid, intuitive decisions rather than
perform the more difficult analyses’ (Pierce, 1996)
– This suggests that decision aids can support decision making if
they can help ‘counteract’ this tendency to take ‘shortcuts’ (or
satisficing) tendency
especially when the decision is important, and when there is enough
time to do the analysis
Tomorrow in lab
Review lab 1, part 1
Turn in reports
Demonstrate quantitative aspects related
to decision making & uncertainty (tutorial)
Introduce Lab 2