Download Recent Developments in the Psychology of Risk

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Risk perception and risk taking:
Psychological perspectives.
J. Richard Eiser
Centre for Research in Social Attitudes
Department of Psychology
University of Sheffield
Risk and probability:
Annual risk of dying from...
•
•
•
•
•
•
•
Smoking 10 cigarettes a day
Influenza
Road accident
Accident at home
Accident at work
Homicide
Railway accident
1 in 200
1 in 5000
1 in 8000
1 in 26000
1 in 43500
1 in 100000
1 in 500000
Estimating probability.
• Some risks are underestimated, some are
overestimated, and some are unknown.
• Statistical reasoning is difficult, so people
rely on ‘rules of thumb’, e.g. frequency of
news reports, personal experience.
• Personal experience can be unrepresentative.
• People often don’t see population statistics
as personally applicable.
Where do risks come from?
• Risk involves uncertainty about the
likelihood of events and the value of their
consequences.
• Risk arises from interactions between
people and their social and physical
environment.
• Risk depends not only on physical
conditions but also on human actions and
decisions.
Risk and uncertainty
Risky decisions involve uncertainty about:
• What will happen (likelihood);
• How good or bad the consequences will be
(value).
• Risk controversies can reflect divergent
values as well as disagreements over facts.
Key psychological processes
• Attitudes – what do we want, like or fear?
• Decisions – how do we choose what to do?
• Learning – how do we come by our
attitudes and beliefs?
• Social influence – how do other people
influence our attitudes, beliefs and
behaviour?
Attitudes
Communicators, advertisers, etc. attempt to
change people’s attitudes by:
• Changing perceived likelihood of
consequences;
• Associating positive or negative value with
objects or activities.
• Such associations may be emotional rather
than factual.
Sunscreen use on Tenerife (%)
Locals
Continentals
British
No sunscreen 40.2
22.2
27.8
SPF = 1 - 4
22.4
16.7
24.1
SPF = 5 – 14
29.0
40.8
40.8
SPF = 15+
8.4
20.4
7.4
Optimism vs.
Sunscreen (r)
Optimism vs.
Exposure (r)
Fair skin vs.
Sunscreen (r)
-
-
negative
negative
negative
positive
-
positive
negative
Decision-making under uncertainty
Danger
?
Cautious
criterion
Risky
criterion
Safety
Diagnosis
Disease present
Disease absent
True positive
Correct diagnosis
Appropriate
treatment
False negative
Incorrect
diagnosis
Failure to treat
Actual disease
status
Disease present
Disease absent
Learning
False positive
Incorrect
diagnosis
Inappropriate
treatment
Not necessarily
Unless
inappropriate
treatment harms
patient, improved
health increases
confidence in
treatment.
True negative
Correct diagnosis
No unnecessary
treatment
Hopefully
Poor outcome
following failure
to treat should
prompt test
reappraisal.
Learning
• We tend to repeat actions, or stick with
beliefs, that have brought us success or
pleasure in the past (‘reinforcement’).
• We tend to avoid activities and objects that
have brought us failure or displeasure.
• Hence, we tend to ‘over-sample’ more
familiar (trusted) activities and objects.
• Expectancies of good or bad outcomes may
be confirmed or contradicted by experience.
Feedback from experience is
typically selective
 Approach behaviour can produce
informative feedback whereas avoidance
behaviour does not.
 If you avoid something you believe to be
dangerous, you will typically not discover
whether your fears were justified.
 This means that ‘risk-averse’ fears and
prejudices may persist because they aren’t
fully tested.
Feedback may be delayed or
inconsistent
• Risk-taking (incorrect approach behaviour)
may also persist because feedback is…
• Delayed – e.g. harmful effects of smoking;
• Inconsistent – e.g. not all cases of
dangerous driving lead to accidents.
The Beanfest Project
Collaborators:
Russell Fazio Natalie Shook
(Ohio State)
Tom Stafford Tony Prescott
(Sheffield)
Beanfest
• Beanfest is a virtual world containing good
and bad beans.
• Good beans give you energy, bad beans
make you lose energy. If you lose too much
energy, or never eat, you die.
• You have to learn which beans are good and
bad, but to do this you have to try them. If
you don’t try them, you’ll never find out.
The beans
•
•
•
•
The beans vary in terms of two dimensions:
Shape, from round to oblong
Number of speckles
Clusters of good and bad beans are located
in different parts of the 2-d ‘space’.
Speckles (Y)
10
9
8
7
6
5
4
3
2
1
Shape (X)
1
2
3
4
5
6
7
8
9
10
Experimental findings
• Participants find the task quite difficult.
• Good beans are learnt less well than bad
beans, some good beans being consistently
avoided as though they are bad.
• Generalisation occurs to untrained beans on
basis of proximity (similarity).
• Participants learn to correct positive but not
negative ‘prejudices’.
“Points” Experiment:
Mean Proportion Correct
as a Function of Feedback and Valence
Test Phase
0.75
0.70
0.65
0.60
0.55
0.50
Contingent Feedback
Positive
Full Feedback
Negative
ACTION S ELECTION
MECHANIS M
1 output unit
(action)
1 state unit
(‘energy’)
noise
1 output unit
(evaluation)
LEARN ING
S YS TEM
3 hidden units
22 input units
11 to encode shape (X axis)
11 to encode speckles (Y axis)
Results of simulations
• Bad beans are learnt better than good beans.
• Some good beans are consistently avoided
as though they are bad.
• Such learning generalises to novel beans
with similar features.
• If connections are ‘seeded’ to produce
specific errors, the network corrects
positive, but not negative, ‘prejudice’.
Implications for decision-making
If decisions are guided by expectancies…
• Choices will be risk-averse (for gains)
where this yields satisfactory outcomes.
• Individuals will experience more positive
than negative outcomes.
• Unknown objects will be assumed to be
more negative than positive.
And in real life..
• If we have sufficient resources and freedom
of action, we may avoid anything we
expect, rightly or wrongly, to be unsafe.
• Most of our experiences will then be happy
and we’ll be pretty pleased with ourselves.
• But many people don’t have the resources
or freedom to control their experiences in
this way.
Social influence
• Where we lack direct personal experience it
often makes sense to imitate, or learn from,
others. But which others?
• We are more likely to trust our friends.
• Our friends will tend to have similar
attitudes and experiences to our own.
• Hence, social networks will tend to
reinforce our existing attitudes.
• Media may amplify or attenuate the
perceived seriousness of hazards.
Trust
• Trust can involve reliance on others both as
controllers of risks and as informants about
the extent of any risk.
• Trust can depend on implicit estimates of
the others’ competence, partiality and
honesty.
• If ‘experts’ are seen as having a vested
interest, this may undermine trust.
Why risk perceptions resist change
• We (fail to) test our prior beliefs, by looking
for confirmation rather than falsification.
• We rely on those we like and trust for
information, but such people are more likely
to share and confirm our prior beliefs.
• We all process information selectively, but
typically lack insight into how selective
we’re being.