Quantifying Human Health Risks from Virginiamycin (VM) Use in

Descriptive decision theory
Neuroeconomics, psychology, behavior
Making risky decisions
• Descriptive models: How do we judge
risks and decide what to do?
– What do we worry about?
• Prescriptive models: How should we
judge risks and decide what to do?
– What should we worry about?
Descriptive decision theories
• Neuroeconomics
– Brain imaging, interacting subsystems for
evaluating prospects
• Psychology
– Psychometric paradigm
– Heuristics & biases (Tversky & Kahneman)
• Anthropology/sociology
– Cultural theory
• Social amplification
Neuroeconomics perspectives
The brain perceives, evaluates
and decides
• Brain has different systems for
– valuing positive and negative changes
– delayed and uncertain rewards
– Making risk and value trade-offs
– Taking dangerous risks
• Fast (emotional) and slow (cognitive)
decision processes
• Reinforcement learning
• Science of good and bad risk judgments
Functional magnetic resonance imaging
(fMRI) shows the brain working
http://themindperspective.wordpress.com/category/science/page/3/
http://farm1.static.flickr.com/26/66033055_75befbfc65.jpg
The brain is a network of processes
that interact (and compete) to make
risky decisions
• Nucleus accumbens anticipates
financial gains
• Medial prefrontal cortex
evaluates gains
• Insula anticipates financial losses
• Prefrontal and cingulate cortex
evaluate expected reward, VOI,
cost, and actual outcome
• Amygdala and orbitofrontal
cortex respond to unknown
probabilities, help in learning
http://thesituationist.files.wordpress.com/2007/09/risky-decisions-on-the-brain.jpg
Neurotransmitters connect decision
subsystems
• Coordination
• Conflict
• Weighting
• Resolution
http://www.basisonline.org/images/2008/02/14/34_copy.jpg
Hormones and neurotransmitters
affect risk attitude & behaviors
Testosterone and
economic return.
Average profit and
loss (P&L) on days
when subjects had
11:00 a.m.
testosterone above
(High) and below
(Low) median.
©2008 by National Academy of Sciences
Coates J M, Herbert J PNAS 2008;105:6167-6172
http://www.economist.com/sciencetechnology/displaystory.cfm?story_id=14301951
Cognitive decision models are
independent of “wet-ware”
http://www.stat.columbia.edu/~cook/movabletype/mlm/krantzmap.png
Summary on neuroeconomics
• Large, new, rapidly growing field
• Shows how different parts of the brain
interact via neurotransmitters to affect risk
judgments, risk-aversion (anterior insula)
and risk-taking decisions
• Intersects with evolution of risk-taking
attitudes and “strategies” in populations
• Overlaps psychology of learning adaptive
behaviors in uncertain environments
• Too new to yield a “science of choice” yet
– but provides many useful insights
• Risky choices have real consequences
Summary on neuroeconomics
• Large, new, rapidly growing field
• Shows how different parts of the brain
interact via neurotransmitters to affect risk
judgments, risk-aversion (anterior insula)
and risk-taking decisions
• Intersects with evolution of risk-taking
attitudes and “strategies” in populations
• Overlaps psychology of learning adaptive
behaviors in uncertain environments
• Too new to yield a “science of choice” yet
– but provides many useful insights
• Risky choices have real consequences
http://superconductor.voltage.com/risk
Psychometrics paradigm
Descriptive decision theories
• Neuroeconomics
– Brain imaging, interacting subsystems for
evaluating prospects
• Psychology
– Psychometrics paradigm
– Heuristics & biases (Tversky & Kahneman)
• Anthropology/sociology
– Cultural theory
• Social amplification
Many risk perceptions are
inaccurate
• We often overweight rare events
• Under-weight
common ones
http://www.icsu-scope.org/downloadpubs/scope27/images/fig16.5.gif
Many risk perceptions are
inaccurate
Yet, risk
perceptions
drive societal
resource
allocations
http://www.icsu-scope.org/downloadpubs/scope27/images/fig16.5.gif
Why are risk perceptions and riskbenefit judgments so often inaccurate?
Our judgments are…
• Sensitive to irrelevant information
• Insensitive to relevant information
• Biased
– Illusion of control, self-serving biases
• Sensitive to details of framing
• Reflect wishful thinking
• Reflect irrational endowment effects,
mental accounting effects
• Inconsistent over time and contexts
http://brainandspine.titololawoffice.com/Skateboard.jpg
Psychometric paradigm: 18 factors
that affect risk perceptions (Slovic)
Catastrophic
potential
Familiarity
Understanding
Personal Control
Voluntariness
Many fatalities per event; LPHC (low
probability, high consequence) events
Unfamiliar or novel risks
Poorly understood activity or technology
Example: Driving car vs. taking plane
Driving car vs. nuclear power
Children
Future generations
Identifiable victims
Dread
Trapped miners vs. anonymous drivers
Fear of the unknown/unfamiliar
18 factors that affect risk
perceptions (Slovic)
Trust
Media attention
Accident history
Equity
Some receive benefits, others bear risks
Unclear benefits
Irreversible
Personal
Man-made
Near-term
Could affect me!
Risk perceptions can be
clustered in various ways
Dreadknowledge
clustering
(Paul Slovic)
http://www.prmia.org/Weblogs/General/DavidKoenig1/2008/01/watching_the_so_1.php
Risk perceptions can be
clustered in various ways
Observablecontrollable
clustering
(Granger
Morgan, 1993)
https://apps.who.int/pcs/risk-assessment-ehc/docs/ehc210_fig_3.jpg
The Affect Heuristic:
Emotional/intuitive evaluation
• A stimulus tends to be perceived as
attractive or repulsive.
– Quick, unconscious classification
• Attractive stimuli tend to be seen as
low-risk, high-benefit; unattractive
stimuli tend to be seen as high-risk,
low-benefit
• Rational (cognitive) assessment may
be significantly different
Preferences drive perceptions (!)
Theory
Reality
Ganzach et al. 2008
journal.sjdm.org/7424/jdm7424.html
Affect heuristic (Slovic et al. 2002)
2 groups of subjects evaluated a scenario in which an airport
must decide whether to spend money to purchase new
equipment, while critics argue money should be spent on other
aspects of airport safety.
Response scale: 0 (would not support at all) to 20 (very strong
support).
• "Saving 150 lives" had mean support of 10.4
• "Saving 98% of 150 lives" had mean support of 13.6.
• Even "Saving 85% of 150 lives" had higher support than simply
"Saving 150 lives”!
Affect heuristic (Finucane et. al. 2000)
Information that increases perceived risk decreases
perceived benefit.
Information that increases perceived benefit
decreases perceived benefit.
Time pressure and sparse information increase
reliance on affect heuristic  increase inverse
relation between perceived risk and perceived
benefit
http://www.singinst.org/upload/cognitive-biases.pdf
Implication: Intuitive and popular
rankings and ratings may lead to
ineffective resource allocations
• Example: If you could spend money to solve
one of these problems, which would benefit
people most/save most lives?
– HIV combination prevention
– CO2 reduction & global warming
– School deworming & nutrition programs
– Rural water supply
– Micronutrient supplements for children
– Malaria prevention
• (Any answers will be somewhat controversial –
but some fairly objective analysis is also
possible.)
Copenhagen consensus
Results 2008
In the Copenhagen Consensus 2008, the solutions for global problems have been ranked
in the following order (here the first 20):[6]
1. Micronutrient supplements for children (vitamin A and zinc)
2. The Doha development agenda
3. Micronutrient fortification (iron and salt iodization)
4. Expanded immunization coverage for children
5. Biofortification
6. Deworming and other nutrition programs at school
7. Lowering the price of schooling
8. Increase and improve girls’ schooling
9. Community-based nutrition promotion
10. Provide support for women’s reproductive role
11. Heart attack acute management
12. Malaria prevention and treatment
13. Tuberculosis case finding and treatment
14. R&D in low-carbon energy technologies
15. Bio-sand filters for household water treatment
16. Rural water supply
17. Conditional cash transfer
18. Peace-keeping in post-conflict situations
http://www.theboywhodeniedwolf.com/chart2.jpg
19. HIV combination prevention
20. Total sanitation campaign
Implication: Intuitive and popular
rankings and ratings may lead to
ineffective resource allocations
• Example: If you could spend money to solve
one of these problems, which would benefit
people most/save most lives?
– Micronutrient supplements for children (1)
– School deworming & nutrition programs (6)
– Malaria prevention (12)
– Rural water supply (16)
– HIV combination prevention (19)
– CO2 reduction & global warming (> 20)
• (Any answers will be somewhat controversial –
but some fairly objective analysis is also
possible.)
Implication: Intuitive and popular
rankings and ratings may lead to
ineffective resource allocations
• Example: If you could spend money to solve
one of these problems, which would benefit
people most/save most lives?
– Micronutrient supplements for children (1)
– School deworming & nutrition programs (6)
– Malaria prevention (12)
– Rural water supply (16)
– HIV combination prevention (19)
– CO2 reduction & global warming (> 20)
• (Any answers will be somewhat controversial –
“Social amplification” affects but some fairly objective analysis is also
possible.)
perceived risks and priorities
https://chavi.org/wysiwyg/images/GHVE-Stakeholders.jpg
Post-2015 Copenhagan
Consensus
http://www.copenhagenconsensus.com/sites/default/files/post2015brochure_m.pdf
Post-2015 Copenhagan
Consensus
http://www.copenhagenconsensus.com/sites/default/files/post2015brochure_m.pdf