Understanding Why the Allais Paradox is Rational (2015) | ||
Suppose
somebody offered you a choice between two different vacations: ·
A
50 percent lottery of winning a three-week European tour. ·
A
100 percent guaranteed one-week tour of Europe. Empirically, the vast
majority of survey takers on this and related questions preferred the
second option – take the one-week tour for sure.
After
that choice is made and fresh on your mind, suppose you get to choose an
additional, follow up tour: ·
A
5 percent lottery of winning a three-week European tour. ·
A
10 percent lottery of winning a one-week European tour. Empirically,
the vast majority of survey takers preferred the first option – a chance
at a longer three-week tour. This
is the basic Allais
Paradox, a famous challenge to existing, broadly used Utility Theory.
Backing
up, Utility Theory is a mathematical economic function that describes how
rational, numerically derived people consistently behave.
If psychology is the study of how people behave, economics is a
subset of psychology that reframes and constrains the vast idiosyncrasies
of human behavior into an incentivized norm.
Using this norm – that people have incentives to do what is best
for them – then people should attempt to maximize their benefits.
Framing this in terms of numerical plots, the average person as the
first moment or location parameter of the population will seek equilibrium
at the peak utility where the derivative of the curve is zero.
i.e. Given a choice, most people would prefer to holdout and
receive 100 dollars rather than 95 or 90 if those were the only possible
utilities. Applying
Utility Theory to the vacations, we can numerically translate the first
vacation thusly: ·
(50%)(3
weeks) = 1.5 weeks expected ·
(100%)(1
week) = 1 week expected Making
the vast majority selecting the 1-week guaranteed tour as incorrectly
non-optimal. The second
vacation also translates into: ·
(5%)(3
weeks) = 0.15 weeks expected ·
(10%)(1
week) = 0.1 week expected Making
the vast majority selecting the 3-week lottery as correctly optimal. The
empirical results do not support this Utility Theory for two reasons.
The first vacation results are clearly inconsistent with empirical
findings. Also, the people
are not consistent with themselves by switching on the second vacation
preferences. This
finding partly inspired later work on Prospect
Theory, which adds more variables to better fit the utility curve to
the empirical findings. Namely,
Prospect Theory applies a non-linear preference curve such that the
difference between a 100% probability and a 99% probability is far greater
than the difference between a 50% and a 49% one.
The preferences at the extremes (near 0% or near 100%) shift
exponentially while the preferences at the center (from 10-90%) shift only
slightly. Translated into the
vacations, Prospect Theory would state something like the following: First
vacation choice ·
(1)(50%)(3
weeks) = 1.5 weeks equivalent ·
(2)(100%)(1
week) = 2 weeks equivalent i.e.
“One in the hand is worth two in the bush” Second
vacation choice ·
(0.9)(5%)(3
weeks) = 1.35 weeks equivalent ·
(1)(10%)(1
week) = 1 week equivalent Making
the empirical choices near optimal under Prospect Theory.
It
sounds nice to have human nature so neatly wrapped up in a series of
numbers. Only except that according to Kahneman himself, economists
and decision scientists still often use the basic Utility Theory.
Utility Theory is simpler and acts well as a rule of thumb,
especially in an uncertain world where the majority of probabilities are
near the center rather than at 0% or 100%.
Viewing the equations above, Utility Theory has only 2 factors.
Prospect Theory translates into 3.
In
basic modeling, the presence of an additional term or factor requires
sufficient justification. On
a favored model, additional terms or factors require massive
justification. Utility is a
favored model of rationality. It relies on the assumption that the mean of a population as
the first moment taken en masse is rational.
While it may not capture all behavior, there is an (overly?)
simplistic underlying rationality model that derives into a 2-factor
equation. Prospect
Theory is an irrationality patch to a model of rationality.
It relies on the assumption that the people en masse are not
rational, that a non-linear term must modify the equivalency to fit the
empirical behavior. While it
captures the empirical behavior better, there is technically a lack of an
underlying model that derives the 3rd factor.
How does one model irrationality? This
is not the fault of Prospect Theory.
It is easy to dismiss this as splitting hairs, or as economists
playing with numerical models of banking and lottery and vacations.
But we should not. This
is the fault of neuroscience. When
Prospect Theory dives into the “Fear and Greed” motivators of
economics and terms it, “Loss Aversion,” it references the amygdala as
the locus of quick thinking fear. The
amygdala
is located in the medial temporal lobes of the brain as part of the limbic
system. Simply put,
activating the amygdala generates a fear sensation that is so rapid as to
be essentially subconscious. Ever
hear about those experiments where a fearful image shows up on a movie
screen for a split second and the viewers have no recollection of seeing
the image, yet have elevated heart rates and elevated stress hormones?
That is the amygdala speaking.
This is why Kahneman refers to it in his book, Thinking
Fast and Slow. This is
part of the fast thinking, fast being the irrational, heuristic, biased
judgment portion. So
why do we have amygdalas? Is
it an evolutionary throw-back to our limbic, lizard-like ancestors?
Is this something wrongly mal-adapted to modern life and thus
something to fight against (i.e. suppress our fast thinking amygdalas) so
we can better fit rationality? This
is exactly the track economists take, including Kahneman, the Chartered
Financial Analysts association, and others.
This
is what happens when neuroscience is playing catch up in the supporting
role. Taking a step back,
these vacation choice results are natural human behavior.
These are known, empirical findings on freely behaving humans in
vivo. This is a treasure
trove of data. Why are there
not any neuroscience leads? Why
is this still relegated to financial economics?
This is not irrational behavior.
This is human behavior of Social Neuroscience.
From
the perspective Social Neuroscience work, the issue is not that the
2-factor Utility Theory calculates one result and the 3-factor Prospect
Theory calculates another to fit the empirical findings.
The issue is that the vacation choices are not static and
independent even though Utility Theory and Prospect Theory treat them as
if they were. In
a Social Neuroscience setting, nothing is independent.
Everything feeds back into itself contextually, temporally and
spatially. At each step in
time, the two spatial choices define each other: a 50% lottery of winning
a European tour sounds great, until we hear the alternative of a
guaranteed tour elsewhere. And
at each spatial choice, the future and past temporal results define the
choices. Any tour or chance
thereof sounds great, until we hear or imagine someone else getting a
better tour. This contextual
feedback violates mathematical principles (Glimcher,
2003) and cannot be modeled
as such with economic mathematics. There,
individuals are i.i.d., AKA independent and identically distributed.
Each individual acts in isolation to maximize their benefits,
creating a population mean that maximizes its benefits.
But in Social Neuroscience, this feedback is a core essential.
If the ecosystem can be designed appropriately, the feedback will
factor itself out without leading to runaway additive feedback activity.
In this factored out case, the individuals still seek to maximize
their benefits and behave rationally under the 2-factor Utility Theory
framework. The only
difference is in what the individuals are attempting to maximize.
In
Social Neuroscience, when translated into numerical terms, the individuals
are not trying to get 100 dollars. They
are trying to get 100 followers. In
Social Neuroscience, the primary brain region activity of focus is not the
amygdala of fear. It is the
prefrontal cortex of anticipation and decision-making. In
Social Neuroscience, the translation of the vacation options might be
something like: First
vacation choice ·
(50%)(3
weeks) = unnecessary risk, looking stupid in comparison ·
(100%)(1
week) = keeping with the Jones Second
vacation choice ·
(5%)(3
weeks) = necessary risk, hero-like in comparison ·
(10%)(1
week) = keeping with the Jones The
differentiating concept is that both the Utility Theory and Prospect
Theory operate on maximizing the tour duration whereas the Social
Neuroscience operates on maximizing the relationships with all potential
observers and peers. It requires all individuals to anticipate what everyone else
would do and project forward to anticipate how their actions will be
perceived. Technically, under
this paradigm, people are rational and will rationally loosely follow
Utility Theory while still being consistent with empirical findings. The only difference is in defining what is the utility.
|