Transcript Evolutionary Explanations for `irrationality`

Evolutionary Explanations for
‘Irrationality’
Leeann Breeze
“In formal logic, a contradiction is the
signal of defeat, but in the evolution of
real knowledge it marks the first step
in progress toward a victory.”
- -Alfred North Whitehead
EVOLUTIONARY PSYCHOLOGY
& reasoning today
All humans today carry cognitive traits that served to help our
ancestors survive and reproduce
in prehistoric environments.
Today’s world is much
DIFFERENT than the world
in which our ancestors
lived and evolved
…So the traits we observe
today may have been
valuable in the past,
but some no longer
serve any evolutionary
advantage, given the nature
of modern environments
This includes the psychological strategies we have evolved to use…
EVOLUTIONARY PSYCHOLOGY
& reasoning today
…THUS NON-NORMATIVE REASONING
EXISTS TODAY
Because we evolved cognitive modules that
served for efficiency, reproductive/social
success, and environmental
safety/typicality in prehistoric contexts
So, where Alfred Whitehead is concerned, the contradiction between
Normative and Descriptive theory is a failure for formal logic (which is
hard to argue against) BUT the fact that we evolved to demonstrate
this contradiction because of ecologically sound reasoning marks the
success of “the evolution of real knowledge”
HOW IS IRRATIONALITY ADAPTIVE?
LET’S BEGIN DEMONSTRATING THE
ECOLOGICAL BENEFIT OF IRRATIONAL
BEHAVIOR AND COGNITIVE PROCESSES
BY OUTLINING THE FORCES THAT
GUIDED THE DEVELOPMENT OF HUMAN
COGNITIVE TRAITS, ACCORDING TO
EVOLUTIONARY THEORY
PRINCIPLES OF EVOLUTION
by Natural Selection, from Darwin
1.
2.
3.
4.
Traits show variation
Some variation is heritable
Individuals differ in fitness (the number of
offspring they are able to produce)
A correlation exists between phenotype and
fitness
EVOLUTIONARY
INTERPRETATIONS
A trait’s adaptiveness is determined by its
frequency in the population of interest
An adaptive phenotype will have an advantage
for personal fitness, those who exhibit it will
more frequently survive to reproductive age,
and the trait will be inherited by offspring,
increasing the trait’s frequency in the
population
EVOLUTIONARY TIME
Humans are biological creatures programmed
by evolution to act, think, feel, and learn in
ways that have fostered survival over many
past generations.
Traits we see today exist because they
survived challenges of past environments in
which our ancestors lived
EVOLUTIONARY INTERPRETATIONS
& REASON
Since cognition is not a physical trait, selection
acts upon manifested behaviors that result
from cognitive ability or task construal.
Evolutionary psychology works on the
assumption that cognitive traits we observe
today developed as responses to problems
our ancestors faced over thousands of years
of evolution in prehistoric, savanna-style,
hunter-gatherer, societies that relied on social
interaction to thrive
EVOLUTIONARY PSYCHOLOGY
“Evolved psychological mechanisms are
functional; they function to solve recurrent
adaptive problems that confronted our
ancestors.”
–David Buss interview in Barker, 2006, pp. 69-70
RATIONALITY & EEA
According to hypotheses that reference the
Environment of Evolutionary Adaptiveness(EEA),
our mental modules have structures that are better
adapted to past environments than the present.
WHERE RATIONALITY IS
CONCERNED:
Where do these environmental discrepancies apply?
EVOLUTIONARY INTERPRETATIONS
& RATIONALITY
Because our cognitive modules evolved to serve SURVIVAL AND
REPRODUCTION in the highly social, hunter-gatherer, savannahstyle EEA, our reasoning abilities today do not appear to fit modern
normative theories of logic…resulting in apparent reasoning “errors”
defined as mismatching between descriptive and normative models of logic
EVOLUTION and CHANGES
The rapidly changing technological environment in
which we live makes these previous adaptations
seem even more out-of-date in their modern
context
Because even today, we appear to be designed to
more readily respond to tasks with the influence of:
1.
Typicality of events & Natural Sampling
2.
Social Contexts
3.
Time/Effort-Saving Heuristics
even when these strategies produce obviously
incorrect responses to modern problems
LET’S REVIEW:
WHAT IS RATIONAL?

BARON: anything that helps us achieve
our goals

DAWES: rationality is avoidance of selfcontradiction
 Ascribing to formal (normative) rules of logic
Empirical demonstrations of irrationality

WASON 4-CARD PROBLEM

BAYESIAN INFERENCE

PROBABILITY ESTIMATES
PART I: WASON 4 CARD PROBLEM
Leda Cosmides and John Tooby
The Scenario:
GROUP 1: 4 cards are on a table
There is ONE RULE:
To have a B, there must be 21 or higher on the other side
WASON 4 CARD PROBLEM
GROUP 1: What is the maximum number of
cards you must check to be SURE this rule is
satisfied?
WASON 4 CARD PROBLEM
GROUP 2: 4 cards are on a table
There is ONE RULE:
To have a beer a person must be 21 or older
WASON 4 CARD PROBLEM
GROUP 1: What is the minimum number of
cards you must check to be SURE this rule is
satisfied?
WASON 4 CARD PROBLEM
FINDINGS: although the 2 problems have the same
logical structure, less than 25 percent of college
students can solve the problem for group 1, but
roughly 75 percent of college students answer the
problem of group 2 correctly
After re-designing the problem to eliminate
issues of familiarity, Cosmides and Tooby conclude
that we seem to be predisposed to more easily solve
the problems that involve
CHEAT DETECTION
WASON 4 CARD PROBLEM
& irrationality
Therefore, people seem to violate Dawe’s
definition of rationality by failing to be
consistent when the 2 problems have the
same logical structure
People also violate the 3rd law of rationality
when they are placed with problems like
group 1 by failing to follow normative models
WASON 4 CARD PROBLEM
& evolutionary theory
Robert Trivers, evolutionary psychologist, has
argued that reciprocal altruism is crucial to
the social evolution of our species.
Additionally, reciprocity can only be
spread if non-reciprocators are punished
WASON 4 CARD PROBLEM
& evolutionary theory
In light of Triver’s theory of reciprocal altruism,
Cosmides and Tooby interpret their findings
as being indicative of an evolved mental
capacity for recognizing when some one has
cheated by violating a SOCIAL CONTRACT
WASON 4 CARD PROBLEM
& cheat detection
Evolutionary strategy holds that individuals are
controlled by behaviors that will serve to maximize
the success of their OWN genes
THUS THE BEST STRATEGY WOULD BE TO
CHEAT (getting all possible gains for oneself & profit
from the good nature of others) AND NEVER
RECIPROCATE
WASON 4 CARD PROBLEM
& reciprocal altruism
THEORY: altruism/reciprocity gene and cheater gene
are both FREQUENCY DEPENDENT.
Because if there were too many cheats, competition
would override. But, eventually a random mutation
for a set of ‘altruist genes’ in the cheater population
would begin to have an advantage. Likewise, in an
entirely altruist population, the best strategy is to be
a cheat. and in a mixed population the best strategy
is to be an altruist with cheater-detection, share with
other altruists and punish cheats. So cheaters and
alrtuists hold a balance in our population…
WASON 4 CARD PROBLEM
& cheat detection
…and in a mixed population the best strategy is to be an altruist
with cheater-detection, share with other altruists and punish
cheats.
THUS, WE HAVE EVOLVED TO HAVE PREDISPOSITION TO
GIVE THE NORMATIVELY CORRECT ANSWER TO
SCENARIOS INVOLVING CHEAT DETECTION. AN
APOLOGIST/EVOLUTIONARY PSYCHOLOGIST WOULD SAY
THE REASON FOR THE NON-NORMATIVE RESPONSE TO
THE ABSTRACT PROBLEM IS BECAUSE WE HAVE NOT
EVOLVED IN ENVIRONMENTS THAT PROMOTE THE
DEVELOPMENT OF THE RIGHT EQUIPMENT TO INTERPRET
THE PROBLEM IN A WAY THAT ALLOWS US TO ANSWER IT
CORRECTLY.
PART II: PROBABILITY ESTIMATES
PROBABILITY ESTIMATES
THE LINDA PROBLEM
Linda is 31, single, outspoken, and very bright. She
majored in philosophy. As a student, she was
deeply concerned with issues of discrimination and
social justice, and participated in anti-nuclear and
anti-war demonstrations. .
What happened to Linda? Rank order the following
possible outcomes:
(a) Linda failed to graduate from college
(b) Linda works as a bank teller
(c) Linda works for Green Peace
(d) Linda works as a bank teller and is active in the
feminist movement
PROBABILITY ESTIMATES
The probability that Linda is a bank teller must
be at least as large as the probability that
Linda is a bank teller and active in the
feminist movement, by shear odds
occurrence of ONE event is much more
likely than the combined occurrence of TWO
events
PROBABILITY ESTIMATES
& The Linda Problem
WHY THE ERROR?
Evolutionarily, people have adapted to
assume continuity in the environment. This
seems to have had a consequential effect on
the human affinity for narrative.
We adopt a story of Linda from the snippet of
info, and continue it in our estimates of
likelihood for her future narrative.
PROBABILITY ESTIMATES
& The Linda Problem
This is adaptive because it served to give us
appropriate responses to social environments.
According to Geoffrey Miller, it is adaptive to
assume people are generally consistent, because it
serves our “cheat detection” and “trustworthy mate”
concepts, helping us to better deem who is a safe
reproductive partner—improving our reproductive
success.
So, we may not answer the normative answer to the
Linda problem, statistically speaking…but, we are
answering with the most ecologically-appropriate
response. (Panglossian)
PROBABILITY ESTIMATES
& percentages vs. frequencies
PROBLEM 1: You are a gynecologist who conducts
breast cancer screening in your region using
mammography.
The probability that a woman in this region has
breast cancer is 1%.
If a woman has breast cancer, the probability she
tests positive is 80% (sensitivity).
If she does not have breast cancer, the probability
she tests positive is 9.6% (false positive rate).
A woman tests positive. What is the probability that
she has breast cancer?
PROBABILITY ESTIMATES
& percentages vs. frequencies
PROBLEM 2: You are an experienced physician in a
preliterate society. You have no books or surveys,
only your accumulated experience. A severe
disease is plaguing your people. You have
discovered a symptom that signals the disease, but
not with certainty. Over the years you have seen
many people & most don’t have the disease. Of
those who did have the disease, 8 had the
symptom. Of those who did not have the disease,
95 had the symptom. Now you meet a patient who
has the symptom. What is the chance he has the
disease?
PROBABILITY ESTIMATES
& percentages vs. frequencies
Which was easier to solve?
IN PROBLEM 2, the solution is simple. Total
people=8+95=103. of that 103, only 8 had the
disease—thus the likelihood of a person coming in
and having the disease is 8 out of 103=VERY
LOW(7.8 percent)
PROBLEM 1: I will go into detail on HOW to solve
problem 1 in the next section. For now, know the
answer here too is 7.8%, and more math is required.
Physicians who typically solve problem 2 correctly give
estimates of problem 1 of roughly 70-80% -- nearly
10 times too high!
PROBABILITY ESTIMATES
& percentages vs. frequencies
Why is the normatively correct answer only intuitive in problem
version 2?
The difference between the problems is the use of percentages in
version 1 and frequencies in version 2
In fact, Cosmides and Tooby (1996) found that when they converted
relevant info from probability to frequency formats in an
experiment, their subjects’ performance improved in parallel
SO WHY DID WE ADAPT TO PREFER FLAT RATES INSTEAD
OF PERCENTAGES?
PROBABILITY ESTIMATES
& percentages
Why frequency preference adaptive?
1. Probabilities and percentages were not an everyday
encounter until the 20th century
2. Formal percentages began as scientific notation
during the 19th century
3. Mathematical probability arose in the mid-17th
century
 Thus, the Environment of Evolutionary
Adaptiveness didn’t have selection pressures
involving these mathematical structures BECAUSE
THEY DID NOT EXIST IN THE EEA.
PROBABILITY ESTIMATES
& percentages
Instead, the EEA built our probability estimates
on naturally occurring phenomena.
Therefore, we base our conclusions on
NATURAL SAMPLING– taking census of
events in the environment and
judging likelihoods based
on past encounters
PROBABILITY ESTIMATES
& percentages
Using natural sampling, we set an event
counter each time some event occurs.
Humans seem to spontaneously count
events, and because this is so automatic to
us, our cognitive processing is more
conducive to problems that suit this type of
format.
So, as the apologists say, we are doing the
best we can with the equipment we have
evolved.
PART III: SOLVING PROBLEM # 1:
BAYESIAN INFERENCE
To find the normative answer to problem 1, we
need to use BAYESIAN INFERENCE.
This is a formula for using base rates,
likelihoods/probabilities of 2 events to come
to a final likelihood estimate for one
occurrence, given some evidence.
BAYESIAN INFERENCE
Suppose we have two events, C and T, with
probabilities P(c) and P(t)
There are two conditional probabilities, P(U|K) and
P(c|t)
We define P(c|t) = P(c)*P(t/c) / P(c)*P(t/c)+
P(not c)*P(t/not c)
This tells us how to go from one conditional probability
to the other
If we know P(c|t), P(c), and P(t), we can calculate
P(t|c)
** assume c is the unknown state (a hypothesis that the patient has cancer)
t is the known information (i.e., evidence a positive on the mammogram )
I.e., we use t to update our probability of c
BAYESIAN INFERENCE
By solving the formula with the given values, we can
reason that the probability of cancer is .078=7.8%
BUT THIS IS NOT HOW WE REASON…
WHY?
- fast and fruegal heuristics are more evolutionarily
adaptive than normative reasoning skills
- We haven’t evolved the capacity to be sensitive to
base rates, in the way that we would need to be to if
we had intuitive guide towards using bayes’ theorem
BAYES THEOREM
why fast and frugal heuristics?
Gigerenzer points out that we use our
impression of what is representative or what
is more familiar to us in order to solve
problems, even when that is not going to
produce normatively correct responses.
This is because ‘FAST AND FRUGAL
HEURISTICS’ are much more effective at
dolving real-world problems quickly with
minimum information
BAYES’ THEOREM
heuristics/evolution
when you look at it from the point of view of evolution, this makes
sense. The adaptive value of saving in the EEa time was very high.
Using utility theories, Bayes theorem, and doing the math to find
the probability of attack from a wild animal given the evidence that
you see him approaching quickly, but you are unsure of how long it
has been since he has eaten could cost you your life. It is better to
use a HEURISTIC—and err on the side of saftey: an overgeneralized false positive is much less detrimental in these
circumstances than a false negative. So, it is suggested that we have
adapted to have less sensitivity to the occurrence of false-negative
rates and we hone in on false positives.
BAYES’ THEOREM
& base rate neglect
TVERSKY AND KAHNEMAN consider the breast
cancer example to include a type of cognitive bias
called BASE RATE NEGLECT
In this specific example, the overall rarity of breast
cancer is being ignored.
Gigerenzer references back to our bias for frequency
data as to why we may neglect base rates: because
in the EEA, a concept such as base rates would not
have existed. Our minds have evolved algorithms
that can only work on the sort of input that would’ve
been available in the EEA, so input such as base
rates are commonly ignored.
EVOLUTION AND RATIONALITY
Our cognitive processes were designed by selection to solve
problems our ancestors faced in the EEA
Cognitive errors arise from rules made based on typicality & natural
sampling that do not fit probability scenarios and contemporary
mathematics
Other cognitive errors arise from our predisposition to favor
reasoning that promotes sociality, rather than normative logic
Additional problems occur because of our tendency to conserve
effort and time by using heuristics
EVOLUTIONARY THEORISTS ACT AS APOLOGISTS
CLAIMING THAT ALL OF THESE ERRORS ARE THE RESULT OF
MAKING THE BEST USE OF OUR COGNITIVE CAPACITIES
GIVEN OUR LIMITATIONS
EVOLUTION AND
RATIONALITY
the end