Who is the postivist?

Tom DiBenedetto tdib at UMICH.EDU
Mon Dec 8 14:41:24 CST 1997

James Francis Lyons-Weiler wrote:

>       So, let me get this right - if I select a page at random
>       from any book, and go the 14th line and find the third
>       letter of the first word, and if it is a vowel, I will
>       encode a similarity as a synapomorphy, if not then
>       I discard it as not a good hypothesis of homology,
>       then I'm doing Hennigian cladistics - simply because
>       it's explicit?  There is more to it -

gee James, dont I write enough? Do I have to include the textbook
with every point I try to make? Yes, there is more to cladistics than
being explicit. I was addressing your point that cladists claim that
our method is anti-authoritarian, I said that it was not the use of
parsimony per se that makes it so, but rather its transparency. I did
not intend to say that transparency is all that cladistics is.

>        Hennig had a model of what evolution does to
>       character distributions - and he apparently consistently
>       felt confident that evolution rarely confuses us, but more
>       often illuminates.  That would be a very nice reality to
>       exist in.

Hennig accepted that evolution is the explanation for the general
overall pattern of similarities and differences amongst taxa. If it
isnt, then one cannot reconstruct phylogeny from the evidence of
character distributions, and we are all out of work.

>       Every time you state that parsimony renders the best
>       hypothesis of organismal relationships and stories of
>       homology, the question of accuracy is not far behind.

Not necessarily. Parsimony approaches makes use of all evidence from
character distributions. As such, it is, to my mind, an obviously
better method than those which rely on modelling a particular class
of evidence. This is so even if you take the position that accuracy
is illusory; that finding correlations in the data is all that we can
ever do.

>       And there are some fairly well known conditions in which
>       other methods outperform parsimony.

If you know the operative model in advance,,,,which you cant.

>       Parsimony outperforms
>       itself when the data are subjected to the critical examination
>       for long edges, for instance.

Elimination of non-historical "evidence" is not alien to parsimony
approaches. Long branch problems are criticisms of the data, not the
method. Show me good reason to conclude that a certain class of
evidence is deceptive, that it is not really homologous even though
similar (or chemically identical), and I wont propose a hypothesis of
homology on the basis of that evidence. If you claim that ALL
evidence for a particular organism is the result of convergence, then
you are claiming that the phylogeny of this critter is unapproachable
by any method.
>>>      others have helped to finish the
>> >       job to move phylogenetic science - beyond parsimony.

>> Ah yes, let us reject some of our homology hypotheses even when there
>> is no reason to do so! What a wonderful idea! It should prove to
>> everyone that we are passionate falsificationists!

>       I'm not sure what you mean by rejecting hypotheses without
>       reason.

That is what a preference for non-parsimonious solutions entails. If
that is not what you meant by going "beyond parsimony", then
excuuuuse me.
>        The real
>       source of "confidence" is not the number of unfettered
>       synapomorphy estimates, but rather the number we find
>       for a group beyond which we expect by chance alone.
>        If we reject the hypothesis through
>       significance testing, then we have seen the type of evidence
>       to the contrary that results from a justified, strong
>       test statement.

I dont think this justifies rejecting the hypothesis; it remains
preferable to all others, All you are saying is that our best
hypothesis is weak. Fine,,,,but I still will prefer it to all others
until I can do better.

>         Hypotheses
>       of monophyly exist independently of the evidence used
>       to test them.  For instance, I might pose the conjecture
>       that primates are monophyletic.  The hypothesis
>       is there, and evidence is waiting to be collected
>       as test statements.

This is true in the abstract, but given that there are enormous
number of potential phylogenies, we usually hold off proposing a
hypothesis until the evidence from at least one character
distribution indicates a particular group.

>> A hypothesis
>> of monophyly (a character) is tested against all other hypotheses of
>> monophyly (characters); the test being the expectation of congruence.

>       This is where I differ - the past is so vast, complex
>       processes about - corroboration should come through
>       congruence for the very opposition reason - that we
>       don't expect it.

gee James, now thats a problem. Our little algorithm wont ever put
groups together for which there is no evidence. And since evidence
tends to make us think that a set of relationships might be true,
then we tend to expect that result. I guess you would be happy with a
group which miraculously emerged from an analysis even though you
never encounteted any evidence for it along the way? Would this be a
group you would have extra special confidence in?
expect; hence the null hypothesis.

>Eldredge and Cracraft also presented the argument that most of the
>misunderstanding between individuals who assert that cladograms can at
>once be constructed and tested with parsimony and those who claim that
>they cannot is entirely semantic.  They argue that the use of the term
>"falsify" can have two meanings in hypothetico-deductive science, and
>that the interpretation they (and others) wished to convey is that
>homoplasies do not absolutely falsify trees as hypotheses (where to
>falsify is equated with to disprove, to cause the rejection of, or to
>refute totally), but rather that they are used to falsify  trees only to
>the extent that the hypothesis is made weaker.  This defense was later
>recapitulated by Farris (1983) and Ax(1987).

I guess I differ. I think that discussion of falsification of trees
is misplaced. We do not test, or falsify or support trees as a whole.
A tree (cladogram) is an epiphenomenon; it is the logical ordering of
grouping hypotheses. We test, falsify or support grouping hypotheses.
All of our science is carried out at the level of grouping
hypotheses, hypotheses of monophyly. The tree of Animalia may have at
its root a grouping of all animals. THis does not specify the fully
resolved tree; it merely distinguishes animals from all other life.
The same goes with every node in the tree. The cladogram is the
ordering of these hypotheses. THere are no tests for the tree as a
whole (except peeking into god's blueprints); there are only tests
for individual grouping hypotheses. If you falsify one, then the
cladogram will be different, because you will have a new set of
grouping hypotheses to logically order

>As Eldrgedge's and
>Cracraft's  use the term falsification in the weak sense suggests,
>inferences that rely on parsimony alone are seated squarely in the realm
>of inductivism, not hypothetico-deductivism.

And that is why I disagree with this. A cladogram is not an
induction; it does not go from evidence to hypothesis - it is the
ordering of hypotheses. Once again, all of the science of systematics
takes place at the level of grouping hypotheses, not at the level of
the ordered set of hypotheses.

> The identification of a
>synapomorphy  based on its transformation on a tree that was selected, in
>part,  because it was not a homoplasy,  is clearly circular.

I dont think it is circular AT ALL. The synapomorphy was/is a
hypothesis of homology/monophyly. It was selected because it passed
the test of congruence with other hypotheses.

> The
>inductive steps include the assumption that a tree of minimum length
>provides a picture of the more general history of diversification, and
>that specific conditions that require few steps applies generally to
>every history of diversification.

No. The tree of minimum length is the efficient ordering of the
consitutent hypotheses of monophyly. That is all.

>       I can't speak for E& C's current position on this - I have
>       encountered numerous authors who will admit that after a
>       decade or so, they can't agree with their past positions.
>       (That's commendable).

Shall we have a beer together in 2007?  :)

>>       Parsimony is not an inductive inference;

>       We generalize from a few synapomorphies to the history
>       of the entire organisms - thai is induction.

Hmmm, lets see,,that is a different axis than what I thought you were
referring to (individual groups -> tree).
But no, I dont think this is induction either. First off, if it is,
then every method would be doing the same thing. But I dont think it
is. We are not inferring the existence of the organism from
characters,,,we have an independant and a priori perspective which
tells us that organisms exist and characters are parts of them.
Unless you wish to raise the possibility that there is no correlation
between the history of characters and the history of organisms, then
I dont think there is a character ->organism induction inherent in
parsimony methods.
>> It is important ot have powerful
>> tests, but accepting falsification when it is not indicated is just
>> silly.

>       I'm not sure what you're referring to.

Accepting less than parsimonious results.
>>>> We do not use parsimony as  a test in itself.

>> >       That's like saying that final exams are a test for a subject
>> >       course, but that the exam is not a test.  ??????????

>> Not really.....its like saying that the exam (test of congruence)
>> will be graded (parsimony criterion), and the exam is not the same
>> thing as the principle of grading.

>       Now you're confusing the criterion with the test.

No, they are different. That is my point. That is what I was trying
to say. Parsimony is the critereion, congruence is the test.
Parsimony is not the test. It could only be the test in a sytem in
which we believed evolution to be parsimonious. THen we would be
testing for the "parsimoniousness" of our distributions. We are
testing for the congruence of the hypotheses. PArsimony merely
assures us that we are conducting the test efficiently.
>       How can you say something is a reconstruction when it
>       is probably wrong?

I dont think that our well-corroborated cladograms are probably
wrong. I dont claim that they are necessarily true, but they are the
best we have, given the evidence of character distributions, and our
accumulated knowledge of biology. I dont know what context you are
calculating your probabilities in, such that you can claim that they
are probably wrong.

>       It seems more forthcoming and accurate
>       to refer to the process called "attempts at reconstruction"
>       and estimate.  Reconstruction alone is misleading to the
>       uninitiated.

Ah,,,,so lets initiate 'em. I have no problem qualifying
"reconstruction" with "attempt at...". I do think the distinction
between reconstruction and estimation is illuminating however.

>>  I have
>> always thought that if the stats folks would translate thier max-like
>> trees into an explicit set of conclusions regarding the homology of
>> the various sites, that we might begin to speak the same language.
>> Then we could begin a debate on what makes a legitmate test of
>> homology.

>       I wonder why they haven't?

Because they have no sense of the organism, they have this myopic
view that the answers to phylogeny can be found in genes alone, and
neednt look elsewhere, and that if a method isnt explicilty
statistical, then it isnt rigorous science. I think some of them are
growing out of that though...
>        When long edges exist,
>       we are guaranteed to have the wrong tree...

Not so at all! Unless you are somehow including in your definition of
long edge some criterion which has conclusively demonstrated this.
Once again, true sister taxa can attract each other, this does not
yeild a wrong tree. Nor do I see why long edges MUST lead to
artifactual attraction. I have seen cases in which a "hypothesis" of
long branch attraction has been conclusively falsified; the branches
were long, but they could not have been attracting each other
artifactually. Now, perhaps they were not proposed as "long" under
criterea that you would use, but that is another matter.
>> The improvements in our field will not come from abandoning the
>> preference for the result which represents the logical ordering of
>> those hypotheses which have survived critical testing.
>> Your comments here seem to contradict what you said a few paragraphs
>> above. You claimed there that we would move along the asymptote to
>> truth by further testing homology hypotheses. That is perfectly
>> consistent with parsimony - for parsimony is (one last time) the
>> logical ordering of tested homology hypotheses. Test 'em more,,,sure,
>> but eventually they will be combined in a matrix and submitted to a
>> congruence test to form the overall phylogenetic hypothesis.
>       Don't you see the difference between hypothesis testing
>       and hypothesis formulation?

I think that you are looking at the cladogram as a hypothesis. It
isnt, in the strict sense - it is an ordering of hypotheses. If you
wish to criticize the ordering, then the discussion would follow the
lines of whether a hierarchical ordering is justified. That is
another matter; one in which i guess many of my adversaries would be
forced to take the same side as me.
Let me ask you this. Do you know of, or conceive of any possible
tests for the cladogram as a whole which is not simply a test of the
subordinate grouping hypotheses? In other words, if you feel the need
to reject a particular grouping hypothesis, then I would claim that
you just put
together the new set of acceptable hypotheses to form the new
cladogram,,,,using parsimony of course.

Tom DiBenedetto                 http://www-personal.umich.edu/~tdib/
Fish Division                                   tdib at umich.edu
University of Michigan Museum of Zoology

More information about the Taxacom mailing list