There are three common mistakes made when debating hypotheses.  The first
mistake is assuming the proliferation of endless hypotheses negates the
one being discussed.  Suppose I have hypothesized that a kibble diet
contributes to insulinoma and other dietary disease.  Does offering an
alternative hypothesis show the first hypothesis is wrong?  Absolutely
not!  Neither has been tested, which means neither has been disproved.
But, it goes deeper than this.  The alternative hypothesis has to be
feasible, the assumption it is based on has to be testable, and it has to
be able to explain the observable data.  A hypothesis that is unfeasible,
or cannot be tested is void.  A hypothesis that cannot explain the
observable data has been falsified.
 
A second mistake is the idea that because something is defined as a
hypothesis, it is just another idea.  While all hypotheses are ideas, not
all ideas are hypotheses.  "What ifs" are NOT hypotheses UNLESS they can
explain observable data, have testable implications, and are feasible.
For example, saying a natural diet is better than a kibble diet is a
hypothesis because it is feasible, testable, and can explain observable
data.  Saying 'some other' diet may be better than a natural diet IS NOT
a hypothesis because it is not testable.  A common and related problem is
the suggestion that just because an idea is a hypothesis, it is unproven.
Actually, it hasn't yet been falsified!  More importantly, if something
is a tested hypothesis, it has ALREADY withstood the rigors of
investigation, and is far more than just another idea!  Suggesting a
natural diet is better is not just a hypothesis, but one that has
withstood millions of years of testing via natural selection.  The more a
hypothesis has withstood the testing process, the stronger it becomes.
It would take an overwhelming amount of data to disprove the hypothesis
that a natural diet isn't the best for a ferret.  Just offering
alternative, untested explanations doesn't cut the mustard.
 
A third common mistake in debate is offering alternative hypotheses
based on formal analogies rather than relational ones, especially when
the point being argued against is based on a relational analogy.  For
example, some people suggesting feeding chicken bones to ferrets is
dangerous because dogs have had so many observable problems.  This is
an example of a formal analogy because it is based on inference, not on
observations.  The analogy is very weak, made weaker when it is pointed
out the way dogs eat chicken bones, the anatomy of the dog mouth and
throat, and the sizes of the two species negate many of the problems.
Observable data suggest ferrets CAN safely eat chicken bones, which
easily falsifies the hypothesis.  If a relational analogy for bone eating
was used from a species having the same size and characteristics of a
ferret (polecats, mongeese, civets, wild cats), the conclusions would
have been vastly different, and it would have been able to explain the
observable data (ferrets safely eat bone all the time).  The problem was
in using a formal analogy where a relational one was demanded.  The vast
majority of alternative hypotheses offered on the FML fall in this
category of mistake.
 
The importance analogs to ferrets is profound.  Even though there are
gaps in knowledge about ferrets, there are numerous studies of the same
problems in other animals (including humans).  These studies can be
safely used as relational analogs to build explanations of what is
happening in ferrets.  For example, ad libitum diets have been shown to
decrease lifespan and contribute to an increase in tumors, lymphoma,
autoimmune problems, obesity, and coronary, hepatic, pancreatic, and
gastrointestinal disease in ALL species tested; using the data to suggest
the same is true in ferrets is clearly a relational analogy (or even a
direct homolog).  This makes the evidence extremely difficult to dispute.
 
The problem isn't an example as clear-cut as the ad libitum diet example,
but when we aren't sure if the analogy is of the formal or relational
type.  Pet food makers have used the success of kibbled diets in one
species as a relational analog to feed the same type of diet in ferrets.
The problem is, because of specific dietary adaptations in ferrets, the
analogy should have been a formal one.  In some cases, the use of analogy
might have to be suspended until the exact relationship is confirmed by
independent testing.
 
In those cases where a study in one species can be shown with observable
data to be a relational analogy with the ferret, then the use of the
information is appropriate, and may be invaluable.  Just because the data
was generated in other species isn't an ipso jure admission it may be
faulty; it is up to the opposition to show the analog is inappropriate.
 
Bob C
[Posted in FML issue 3949]