The Evolution of Human Cooperation Discussion

Mark Pagel: How language transformed humanity | TED Talk

CARTA:Tribal Social Instincts and Human Cooperation – YouTube

People lined up to donate blood after mass shooting in Las Vegas
Human Cooperation
Definition
Altruism:
✤
Common definition: doing something nice for
someone else (when you don’t have to)
✤
The biological definition is much stricter: A behavior
that provides a reproductive benefit to another at a
reproductive cost to oneself
✤
Remember that a reproductive cost or benefit does not
necessarily need to be directly related to reproduction.
Food is a reproductive benefit (you can’t reproduce if
you’ve starved to death).
Most cooperation is not altruism
Kin-selected cooperation is not
biological altruism
✤
Reproductive success is
measured at the level of the
genes
✤
If you benefit a relative, even at
a cost to your own
reproduction, you will be
benefitting some of your genes
(that appear also in your kin)
✤
If the behavior actually follows
Hamilton’s rule, then, by
definition, there must be a net
benefit to the actor
Meerkats stand “sentry,” and sound an alert if a
predator is spotted. This is a potentially very costly
behavior that benefits others: it makes the individual
visible and catches the predator’s attention at the same
time as warning other meerkats. The degree to which
meerkats engage in this behavior is directly related to
the average level of relatedness in their group.
Reciprocal altruism is not actually
biological altruism
✤
Even if there is a delay in the
benefit accrued to oneself, the
behavior evolves because it
ultimately benefits the altruist
✤
BUT, it’s risky… especially if
there is a long delay before
reciprocation occurs
✤
Remember that for the strategy
to evolve, it must provide a
reproductive benefit on average,
across evolutionary time: any
particular individual may
experience a cost when using a
specific strategy
These snails are simultaneous hermaphrodites (they have
both male and female reproductive organs). They are
taking turns providing male (cheaper) and female (more
expensive) gametes. There is a risk that one’s trading
partner will leave after fertilizing expensive eggs with
cheap sperm. If that happens too often, the behavior won’t
evolve, because it results in a net loss for the egg donor.
Risking your life to save a stranger IS
biological altruism
✤
Saving a stranger provides no
kin-selected benefits (remember
that cooperation in non-humans
occurs primarily among kin)
✤
The person saved is unlikely to
ever reciprocate (especially if
they don’t know who saved
them)
✤
Human sociality and the
capacity for culture are critical
for the evolution of large-scale
cooperation (which is not
necessarily altruism) and true
biological altruism
Humans regularly cooperate in ways
that other animals can not
✤
Human cooperation can…
‣ be anonymous
‣ favor nonkin
‣ occur without reciprocation
‣ be entirely voluntary
✤
Some of the many contexts of
human cooperation:
‣ Pick up litter
‣ Give money to charity
‣ Leave tip at restaurant
‣ Give blood
‣ Join demonstrations
Hunger and Homelessness Awareness Week
CSUF, November 18, 2009
Mechanisms that could lead to human
cooperation
✤
Extended reciprocity: the ability to reciprocate with long delays
between acts and with different currencies (e.g. I pay you back for
driving me to the airport by buying you a beer)
✤
Indirect reciprocity: the ability to receive indirect benefits from
otherwise altruistic acts (e.g. firefighters may gain social status, be
particularly desirable to the potential mates, or receive material
rewards)
✤
Cultural evolution: the human capacity for culture (next lecture) can
lead to many behaviors that would not be favored by natural
selection on their own
Axelrod’s Tournament
✤
Axelrod’s Tournament was a series of simulated game
play (using the basic Prisoner’s Dilemma) to test what
kinds of cooperative strategies could evolve
✤
Repeat interactions with unknown number of plays
means that cooperation in one round could always
result, with some probability, return cooperation in a
future round
✤
Mathematicians, theoretical biologists and economists,
and others submitted various “strategies” to compete
✤
The “winner” was the strategy that outlasted all other
strategies – i.e., survived and reproduced better
✤
Note: One of the parameters of the contest was that there was a
small degree of “noise,” that is, mistakes could happen
Robert Axelrod
Axelrod’s Tournament
✤
Many strategies were submitted; some quite complicated and
using intricate algorithms
A very simple strategy won: Tit-for-Tat
✤ Cooperate on the first move, then do whatever your opponent
just did. Tit-for-Tat is
✤
‣ Nice – always starts off cooperating
‣ Forgiving – one cooperation is enough for it to forgive past defects
‣ Retaliatory – one defection is enough to turn mean
✤
“Defect” is the only Nash Equilibrium in Prisoner’s Dilemma,
but both “Defect” and “Tit-for-Tat” are Evolutionary Stable
Strategies (ESSs).
Tit-for-Tat’s retaliatory nature has one
disadvantage
Cooperate first
✓ If partner cooperated,
cooperate
✓
✓
If partner defected, defect
If there are errors or
imperfect information, it
can lead to “blood feuds”
✤ Another submitted
strategy, “Generous Tit-forTat,” is robust to error by
specifying “If partner
defects twice, defect”
Retaliation
Mistake
Tit-for-Tat C, C, C, D, C, D, D, D, D, D, D, …
Tit-for-Tat C, C, D, C, D, C, D, D, D, D, D, …
✤
Mistake
Retaliation
Mistake
GTFT
C, C, C, C, C, C, D, C, C, C, C, …
TFT
C, C, D, C, C, C, C, D, C, C, C, …
Mistake
Retaliation
But “Generous Tit-for-Tat” is not an ESS
✓This leads to oscillating strategies
Random
‣ Tit-for-Tat can “invade” a population of Always Defect
‣ Generous Tit-for-Tat can invade Tit-for-Tat
‣ Always Cooperate can invade Generous Tit-for-Tat
‣ Always Defect can invade Always Cooperate
‣ And so on…..
Always Defect
Always Cooperate
Tit-for-Tat
Generous Tit-for-Tat
Preferential assortment may give a
strategy a boost to invade an ESS
✤
When cooperators are rare:
๏ Given random assortment, they will be most likely to pair with
defectors (0), and defectors will be most likely to pair with
other defectors (1), but sometimes with cooperators (5)
๏
✤
If cooperators can identify and preferentially pair with one
another, while avoiding defectors, their payoff will most often
be (3) and that of defectors will most often be (1)
Existing mechanisms can produce preferential assortment, e.g.:
๏ Kin selection
๏ Direct reciprocity
Cognitive abilities required for
extended reciprocity
✤
Ability to recognize and remember
large numbers of individuals
✤
Ability to track and remember
multiple individuals’ actions across
a wide range of behaviors/activities
✤
Ability to track exchanges between
multiple currencies (e.g. food
sharing, alloparent care,
coalitionary alliance, etc.)
✤
Ability to detect and avoid cheaters
Harder than it sounds…
Figure by Dr. Ashleigh S. Griffin
University of Oxford
✤
An example of just one barrier:
cheater detection
✤
Simple! Let’s have everyone
who is a cooperator develop a
phenotypic marker of being a
cooperator, for example, a
green beard
Mechanisms generating indirect fitness benefits for cooperation. (a)
Limited dispersal. If individuals do not move far during their lifetime, then
they will tend to be surrounded by kin (shaded) and hence even
✤
Problem: this can easily be
invaded by a new phenotype
that falsely displays a green
beard without the linked
altruism trait
indiscriminate altruism could be directed primarily towards kin rather
than nonkin. (b) Kin discrimination. For example, if the actor can
remember those individuals it shared a nest with when young and
discriminate these kin (shaded) from nonkin (unshaded) after leaving
the nest, then cooperation can be directed primarily towards genetic
relatives. (c) Greenbeard. If the gene controlling altruism is also
associated with a phenotypic marker, such as a green beard, then
Cognitive mechanisms
✤
If phenotypic markers of cooperation are evolutionarily unstable,
how can we solve the problem of preferential assortment?
✤
By having cognitive mechanisms that are very sensitive to cues of
cheating
✤
This requires energetic investment in building such cognitive
mechanisms, refining them through learning, etc.
✤
The potential benefits of cooperation must outweigh those costs (on
average, across evolutionary time) in order for such mechanisms to be
favored by natural selection
✤
There is evidence that humans have such specialized mechanisms
A
Small
Test
• Imagine the rule:
‣
‣
If a person has a card with a “D” on one
side, then his card must be marked “3” on the
other side
Each card has a person’s letter on one side,
and his number on the other
• Which are the cards you need to turn over
(fewest possible) to verify that the rule
is not being broken?
D
F
3
5
Person 1
Person 2
Person 3
Person 4
If you said you needed to
check the card labeled
“D” and the one labeled
“3”
You were wrong
😔
(But not alone.
That is the most common
answer)
Try another one
• If a person is drinking beer, then he
must be over 21
‣
Each card has the person’s drink on one side,
and his age on the other
• Which card(s) do you need to turn over to
verify that the rule is not being broken?
Soda Beer
Person 1
Person 2
18
25
Person 3
Person 4
One more…
• Among the Bogon people, there is a rule:
‣
If a person wears cancha beads, then she must
have completed the huarango ritual
• Which card(s) do you need to turn over to
verify that the rule is not being broken?
Cancha
beads
No
cancha
beads
Hasn’t
done
huarango
Has
done
huarango
Person 1
Person 2
Person 3
Person 4
Why are some of these
problems easier to solve?
• Each of these problems presents a
logical rule with the SAME FORM: “If
p, then q”
• To find violators of this rule, you
must only check ‘p’ and ‘not-q’
• The first version had no social
context
D
Is this
marked
3?
F
3
No need
to check
No need
to check
5
Does
this
have a
D?
We are much better at solving
this logical task when it
involves a “social contract,”
or rule.
18
25
Over 21?
Drinking
alcohol?
No need
to check
Cancha
beads
No
cancha
beads
No need to
check
Has
done
huarango
Did Huarango?
Hasn’t
done
huarango
Wearing
Soda Beer
No need
to check
beads?
No need to check
Summary
✤
The degree to which humans cooperate, and the types of
cooperation that they perform, are unique in the animal
kingdom
✤
Natural selection rarely favors large scale cooperation because
it is generally susceptible to cheaters
✤
Humans have evolved psychological mechanisms that make
this possible, including:
‣
An aptitude for applying and understanding social rules
governing access to benefits, and for detecting cheaters in the
environment
Case studies of
cooperation
✤
Evidence for extended
reciprocity in human societies
‣ Chapter 17: Effects of illness
and injury on foraging among
the Yora and Shiwiar, by
Lawrence S. Sugiyama and Richard
Chacon
‣ Chapter 18: Reciprocal
altruism in Yanomamö food
exchange, by Raymond Hames
‣ Chapter 19: Reciprocal
altruism and warfare: A case
from the Ecuadorian Amazon,
by John Q. Patton
©2000, Aldine de Gruyter, New York
Pathology Risk (Yora & Shiwiar)
Adaptation and Human Behavior, Chapter 17
✤
Research question: How do we
explain caring for sick or wounded
individuals, at a cost to oneself?
✤
Two known mechanisms:
✤
Kin Selection
‣
✤
But, exogamy results in many
people in communities with few
kin
Reciprocal Altruism
‣
But, those most in need are also
those who are the least likely to be
able to reciprocate (the Banker’s
Paradox)
Ecuador & Peru
Methods
✤
✤
✤
Two populations, each characterized by hunting, horticulture,
foraging, and fishing subsistence patterns
‣
Yora: When obtained, meat is shared widely throughout the village
‣
Shiwiar: Meat is shared only among established reciprocal households
Both populations have similar risks of injury and illness:
‣
Cuts, burns, animal injuries, fractures, snake bites, warfare, disease, infections
‣
Yora experienced multiple serious injuries/illnesses during the study period;
Shiwiar did not
Compared foraging returns under various conditions:
‣
When all hunters are healthy
‣
When an average hunter is injured or ill
‣
When the best hunter is injured or ill
The Yora: Results
‣
The reference line at 45 g / protein
indicates the USRDA (recommended
daily allowance) for protein
‣
Bar 1 shows the average daily protein
return observed during study period,
which included multiple injuries/illnesses
(avg is above USRDA)
‣
Bar 2 shows estimated daily average if
there were not any injuries or illnesses
‣
Bar 3 shows the estimated daily average
when the best hunter is injured or ill
(well below USRDA)
‣
Bar 4 shows estimated daily average if
best hunter AND one average hunter are
injured or ill
Biggest overall cost occurs when
the best hunter is injured or ill
The Shiwiar:
Results
‣
The reference line at 45 g / protein
indicates the USRDA
‣
Bar 1 shows the estimated average daily
protein return observed if an average
hunter were injured/ill (none were during
the study period); above USRDA
‣
Bar 2 shows the observed average daily
protein return during study period (well
above USRDA)
‣
Bar 3 shows the estimated daily average
if the best hunter is injured or ill (well
below USDA recommended)
‣
Bar 4 shows estimated daily average if
best hunter AND one average hunter are
injured or ill
Again, the biggest overall cost
occurs when the best hunter is
injured or ill
Why take care of the
sick or wounded?
✤
It is in other individuals’ best
interest to care for the best
hunter if he is sick or wounded,
but not necessarily to do so for
an average hunter
✤
This provides clear motivation to
compete to be the best hunter
✤
But, how to protect against freeriders (i.e. those who partake of
the hunter’s returns when he’s
well, but don’t take care of him
when he’s sick)?
More generally
✤
Cooperation can occur in the context of social niche differentiation
(i.e. specialization); each member provides irreplaceable benefits
‣
Evaluate one’s own abilities
‣
Identify the important needs of others
‣
Compare one’s own abilities to those of others
‣
Identify the needs that one is most qualified to provide and which are
not well provided by others
‣
Cultivate abilities in that area
‣
Convincingly signal to others that provision of benefits is a consistent
product of living with one
Yanomamö Food Exchange
Adaptation and Human Behavior, Chapter 18
✤
The research question: What is
the nature of exchange in smallscale foraging societies?
✤
This study tested between two
competing models (data can
support one or the other):
‣
Generalized sharing model
(egalitarian exchange)
‣
Reciprocal altruism
Venezuela & Brazil
Brief description of the models
✤
✤
Generalized Sharing Model
‣
“From each according to ability, to each according to need” (Karl Marx)
‣
The relevant criteria for sharing is the recipient’s need
Reciprocal Altruism
‣
Exchanges are strategic in nature and designed to enhance the fitness of
the person sharing
‣
The relevant criteria is the recipient’s ability to reciprocate
Methods
✤
Modified time-allocation study; any time during observations of
individuals someone was observed eating, information was recorded
regarding type of food, amount of food, and who provided food.
Able to quantify the flow of food between households.
✤
Features of food exchange that were assessed:
‣
Intensity of exchange
•
‣
Scope of exchange
•
‣
The percent of a household’s diet that comes from exchange
The number of households from which one receives food
Balance of exchange
•
Is exchange balanced in general within village, or between
specific households?
Predictions (What would you expect to observe
if either model were true?)
✤
✤
Generalized Sharing Model
‣
Intensity: If sharing is according to need, households with greater
numbers of consumers relative to producers will receive more
‣
Scope: Sharing should include all households
‣
Balance: Balance of sharing should be according to relative ability to
produce and relative need within village
Reciprocal Altruism
‣
Intensity: If sharing is based on strategic self-interest, there need be no
relation between household need and the intensity of sharing
‣
Scope: Sharing should occur between established partners
‣
Balance: Sharing should be balanced between exchange dyads, but not
necessarily relative to village
Results
✤
The results supported the Reciprocal Altruism model, and were not
consistent with the Generalized Sharing model
‣
Intensity of sharing:
•
‣
Scope of sharing:
•
‣
Larger households received greater absolute amounts of food, but
did not receive a larger proportion of their diet from exchange
Sharing was restricted to a few established dyads within the village
Balance of sharing:
•
There was no overall balance within the village, but between sharing
partners, exchanges were balanced
Warfare (Achuar & Quichua)
Adaptation and Human Behavior, Chapter 19
✤
Public goods games
‣
✤
When a public good (e.g. a public
water fountain) is equally available to
all, but not all must share in the cost
of providing the benefit, there is
always an incentive to be a “free
rider” (take the benefit without
paying the cost)
Specific research question:
‣ “Why do warriors undertake the
somatically risky behaviors associated
with war when the benefits derived
from the risks they take appear to be a
public good, that is group defense
and the pursuit of coalitional goals?”
Ecuador & Peru
join others from around the world, and you
Reciprocal altruism allows for
segmental loyalties
will be launching the largest aerial battle in
the history of mankind. Mankind, that word
should have new meaning for all of us
✤
“Segmental loyalties” refers to
the fact that what is defined as
our “in-group” can change
based on circumstances
today. We can’t be consumed by our petty
differences any more. We will be united in
our common interest. Perhaps it’s fate that
today is the 4th of July, and you will once
again be fighting for our freedom. Not from
tyranny, oppression, or persecution, but
from annihilation. We’re fighting for our right
to live, to exist. And should we win the day,
the 4th of July will no longer be known as
Independence Day, 1996
an American holiday, but as the day when
The Study
✤
If warfare is based on
reciprocal altruism, we should
expect to see:
‣
Evidence of social contracts
for mutual defense
(hierarchical loyalties)
‣
Evidence of a shared logic for
distributing status within
coalitions
Results: Hierarchical Loyalties
‣
Multidimensional Scaling (MDS): Each
icon represents one man in the village.
The arrangement of the icons represents
the strength of individual’s alliances,
based on interviews and surveys (e.g.
“who would be most likely to help
who)
‣
The top figure shows the entire
community. There are two clear
coalitions, that largely correspond to
the two ethnic groups present in the
community (Quichua and Achuar)
‣
The bottom figure shows only the
Achuar ethnic group. Within that
larger coalition, there are multiple
distinct coalitions
Results: Shared
Logic
In general, men’s status within the community is related
to their perceived warriorship (ability to win a battle)
The figure on the left shows that, when assessing other men’s status, men attribute higher
status to their own group members than to the other group’s members. The figure on the
right show that, when assessing other men’s warriorship, they do do not show the same bias
Interpretation
✤
The fact that men’s assessments of each others’ status exhibit a clear
in-group bias (they assess members of their own group as having
higher status), but their assessments of each others’ warriorship does
not, indicates that there is an implicit recognition of the reciprocal
value of a man’s warriorship
✤
That is, if people underestimated a warrior’s ability, they would not
reciprocate adequately for the service he provides (defending the
village)
✤
Thus, despite the in-group bias in assessing status, when the group is
taken as a whole, there is a strong correlations between warriorship
and social status, that is, warriors are being reciprocated with status
Summary
✤
These three examples provide ethnographic and empirical
evidence that…
‣
people are sensitive to reciprocal relationships
‣
reciprocity occurs across currencies (e.g. warriorship is repaid with
social status)
‣
people adjust their own contributions in order to both reciprocate
the contributions of other and ensure their own value in the
community
Dual Inheritance
Theory
• Culture exists
• Culture evolves
• Culture is an adaptation
• Culture is maladaptive
An example of between-group variation in
non-human primates
• Baboons are considered by
most biologists to be a single
species
• Baboons occupy territories
across most of Africa,
including several disparate
ecologies
• Baboons vary in:
‣ Physical appearance
‣ Exact composition of diet
‣ Social groups
‣ Social behavior
Human variation dwarfs the variation seen among
any other primate groups – humans vary a LOT
Thought experiment
‣ Imagine we transplanted
groups of baboons between
habitats
‣ We allow no contact with
local baboons
‣ What would happen?
‣ Prediction: We expect that
the transplanted baboons
would rapidly develop
behavior similar to local
behavior and social
organization
‣ Variation is produced by the
environment
Thought experiment II
‣ Imagine we transplanted
groups of people between
habitats
‣ We allow no contact with
local people
‣ If culture is important: We
expect that many differences
between the groups will
persist, even though they
share the same environment
‣ This pattern has been
observed in many historical
cases of immigration and
colonization
Human’s capacity for culture, and cumulative
learning, depends on several unique adaptations
• Imitation
• The ability of culture (that is,
norms, beliefs, etc.) to evolve
in a process analogous to
genetic evolution
• Ability to choose behavioral
models based on:
‣ The content of the behavior
‣ The frequency of the behavior
‣ The qualities of the person
doing the behavior
Humans are exceptionally good imitators
• Among non-human primates and
other mammals, social transmission
is common
• Direct imitation is not (more on this
in a moment)
• Most “learning” in other animals is
observational learning
‣ local enhancement
‣ stimulus enhancement
• Each still requires individual
learning, does not allow for the
cumulation of ideas
Studies show that human babies and apes learn
very differently
• Babies and chimps observed a
researcher manipulate an object in order
to retrieve a treat
• Three conditions: two different modeling
conditions and one with no modeling
• Both human babies and chimps
benefited from having a modeler
• Human babies imitated what the modeler
did
• Chimps got the idea that manipulating
the object gave them a reward, but
figured it out by trial and error
Imitation is necessary for cumulative social
knowledge
• Imitation allows for faster
learning of what is already
known
• We can then use individual
learning to enhance behavior /
knowledge
• Over time, knowledge and/or
skills build upon one another,
and culture changes
• Guided variation: non-random
changes in cultural variants by
individuals that are
subsequently transmitted
This is what makes the process
different from genetic evolution
(Natural Selection). In Natural
Selection, the variation is random.
Culture Evolves
• Like genetic traits, cultural traits
change (evolve) over time. Three
general processes are involved in
cultural change.
• Biased Transmission: change
occurs because people’s psychology
makes them more likely to adopt
some beliefs than others
• Natural Selection: change occurs
because of what happens to people
who hold different beliefs
• Guided Variation: change occurs
because individuals modify existing
beliefs
Mursi woman with lip plate
Content-based bias
• A direct bias
• The tendency to imitate is
based on the content of the
belief or behavior
• If it is easy to ascertain which
variant is more successful,
there will be a bias toward
imitating that variant
Conformism-biased transmission
• A frequency-dependent bias
• Tendency to imitate the
behavior is dependent on its
frequency in the population, not
on its content
• If there is little information about
which variant is most
successful, it is a good strategy
to do what most other people
are doing (they might know
something you don’t)
Prestige-biased transmission
• A model-based bias
• The tendency to imitate is based on
the characteristics of the model
• On average, individuals who have
figured out the best local strategies
will be the most successful
• Because the behaviors leading to
success may be hard to discern,
this bias can lead to a “runaway”
process
• Local leaders are more useful
models than global leaders
Natural Selection of cultural variants
• Natural selection can act on cultural variation by the same principles that it
acts on genetic variation
1. People vary in their beliefs and/or values, and that variation affects
their ability to transmit their beliefs
2. Variation is “heritable”
3. The total number of variants must be limited in some way
• Variants are not limited by environment, but by memory, time, and behavioral
conflicts
• There is no “hidden” variation in cultural variation. Variation that is not
expressed behaviorally (phenotypically) cannot be transmitted
• Transmission is not restricted to parent-offspring; transmission can be from
peers, from children to parents, from an average of many models, etc.
Culture is an adaptation
• Culture looks great! And is clearly a species-typical trait (common to all
humans)
• But why do we have it? And, just as importantly, why do no other species
have it (to the extent that we do)?
• To answer that, we need to think about the potential reproductive costs and
benefits of having this trait to our ancestors.
Benefits
• Cumulative cultural evolution is
essential for human adaptation
• Thought experiment:
‣ Your task is to survive and
raise your kids in the Colorado
Desert of the American
Southwest
‣ You get six months of
supplies… a little time to figure
out how things work
‣ Then all industrialized products
are taken away and you’re on
your own
‣ Will you make it??
Probably Not.
• This area, known as El Camino Del
Diablo, was the best route from Mexico
to California before the railroads.
• But many experienced, desert-wise
pioneers died trying to make the trip
Benefits
• But the Tohono O’odham survived
in this region for thousands of
years
• Able to thrive, grow, raise their
children, all with just a few pounds
of wood, stone and bone
equipment, and a whole lot of
adaptive information that shaped
subsistence strategies:
‣ Semi-nomadic, band social
organization
‣ Foraging plus rainfall horticulture
‣ Specialized technology (tools)
‣ Extensive knowledge of desert
• This information is not contained in genetically
transmitted material
• Rather, evolved psychological mechanisms
allow people to acquire cultural norms,
knowledge, and strategies in an efficient way
Different, culturally evolved adaptations allowed foragers to
occupy a wider ecological range than any other species.
Costs
• Benefit: low cost information
• But… we have to be credulous
(open to many possible cultural
variants)
• Result: maladaptive ideas can
spread
• There are many examples:
‣ Conspicuous consumption
‣ Dangerous hobbies
‣ Drug use
‣ Reduced fertility
‣ Self-sacrificial altruism
Maladaptive variants spread because parents
aren’t the only source of children’s culture
• People in different social roles may be
important in cultural transmission
• People vary in their success at attaining
certain social roles
• This variation is affected by beliefs and
values
• Cultural values that lead to success in
attaining specific social roles will tend to
spread
• Such variants may often be maladaptive
• Famous athletes take horrible risks
• Successful academics give up
reproductive success