Who is the postivist?

[James Francis Lyons-Weiler]

On Sat, 6 Dec 1997, Tom DiBenedetto wrote:

> James Francis Lyons-Weiler wrote:
>
> >        What some see as "modern positivism" can be summarized
> >        as follows:
> >        A misplaced belief that an inference is scientific, and
> >        therefore accurate and precision, simply because it is
> >        based on a formal method.
>
> This is of course, hardly a definition of positivism, nor much help
> to anyone trying to understand what you are referring to.
> Does anyone believe that a method is scientific merely because it is
> "formal"?


        Cladists and others are fond of stating that one of Hennig's
        contributed influences was to standardize the way in which
        phylogenetic and ultimately taxonomic inferences were made.  In
        the dark ages, authoritarianism ruled, and folks created gestalt
        taxonomies.
        Then came cladistics - a formal method.  The question is,
        is that sufficient to make it scientific?  I think from
        your question you might say "No.".  Does formalism make
        phylogenetics _more_ scientific?  Some would say so - at
        the very least it helps it be explicit.  But just because
        one group of people agree that parsimony is the best way
        to perform a phylogenetic estimation problem doesn't make
        the trees correct.  Hennig moved the field to become
        "more scientific", but others have helped to finish the
        job to move phylogenetic science - beyond parsimony.
>
> >     Positivism was adopted by the
> >        Vienna circle, but my studies indicate that they merely
> >        co-apted the British model of empricism; in so doing, they
> >        placed their faith in a single mode of inference, and
> >        became modern positivists.
>
> >From my understanding, positivists can be usefully distinguished from
> Popperian critical rationalists by the former's adherence to a
> verificationist approach to assesing the validity of scientific
> theories, as opposed to Popperian falisificationism. In modern
> phylogenetics, it is (some) cladists who stress the falsificatinist
> approach, whereas the statistical phylogeneticists (those lost in the
> Felsenstein zone) seem to be inherently verificationist.

        Well, in modern phylogenetics as I have come to understand
        it, synapomorphies as read from the tree are taken
        "in support of" hypotheses of homology and monophyly -
        recall all the papers wherein a "List of Synapomorphies"
        are presented?  Each is taken as a confirming instance.
        The more synapomorphies, the better _confirmed_ (others
        disagree and say "corroborated" but it's a loose usage
        of the term).  as such, after much debate, the
        idea emerged that such inferences amount to "weak
        corroboration" but not at all firmly within the realm of
        falsificationism.  And yet you state without reservation
        as if it were entirely agreed upon that parsimony
        corroborates.  As an inductive inference, it simply
        cannot.
>
> >      A positivist
> >        would argue that based on first principle alone, regardless
> >        of how well or poorly it may perform in general, MP provides
> >        the best estimate _in a particular case_.
>
> You are referring to the uses of maximum parsimony in molecular
> "systematics". I think that our discussions have been less than
> efficient because I speak of cladistic parsimony, and that is rooted
> in a very different approach. I have tried to emphasize the
> difference in many of my postings; I am not sure how much success I
> have had.

        I didn't mention molecular anything.  The case has not been
        made that any type of parsimony provides the best estimate in
        any single case.  The difference between molecular data
        and morphological data still eludes me.
>
> Once again, and very briefly, cladistic parsimony utilizes a
> parsimony criterion in the
> implementation of the test of congruence. 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.  ??????????

. (Perhaps James is led to his reference to
> "parsimonious evolution" because of confusion on this very point). We
> begin with hypotheses of character homology. The historical
> interpretation of those homologies (they are "the same" because of
> descent from a common ancestor) implies a test; true historical
> homologies should

        but may not


form a congruent hierarchical pattern with regards
> to their distributions amongst taxa. The test of congruence is
> implemented by combining the homology hypotheses
> to see what sort of a congruent hierarchy exists.

        Appears to exist.  The hierarchy is tree-based, the tree
        is based on the data, so when the data mislead the
        tree, they also mislead the inferences of hierarchy.

. Parsimony is a
> logical principle which underlies the application of this test; less
> than maximally parsimonious solutions are simply inefficient
> implementations of the test; they find less congruence than is
> actually present in the data.

        Less apparent congruence than appears to be in the data
        on the parsimony tree.
.
> The result of a cladistic analysis is a set of character homologies
> which have been corroborated both through the biological tests they
> have endured, and the congruence test. Their logical combination
> represents the sum of phylogenetic evidence from character
> distributions, and our best overall reconstruction

        I prefer "estimation"

of what the
> lineage branching pattern was.
> With James's interest in the "truth", in terms of the "windows to the
> past", I would suggest investigating Popper's notion of
> "verisimilitude"; which is basically the substitute for "truth",
> since truth is unknowable. Verisimilitude is a notion of "closeness
> to truth" and Popper develops a logical calculus for it. I havent had
> time to develop arguments around this concept, but I suspect it will
> be useful for justifying cladistic parsimony approches.

        "Successive approximation" in based on this notion, but
        I wouldn't re-weight my data on the CI or HI of the data on
        the MP tree given the possibility that noise could lead
        me to the wrong tree, to phony and meaningless estimates
        of the informativeness of the data (on the wrong tree, after
        all), even though the fit of the data would seem better.

        As we develop new and more critical tests of hypotheses of
        homology (and there are many that have and well continue
        to be developed), we should be moving farther along the
        asymptote to truth.  For instance, parsimony
        can be mislead by long branch attraction, and
        no model of evolution is needed for that to be true (sparse
        taxon sampling can lead to the same problem and guarantee
        the wrong tree, and we are limited in our taxon sampling
        via extinction), new tests for the identification of
        long edge taxa will prevent errors that would otherwise
        have been made.  Long edge taxa would not be
        found in any parsimony tree- they are hidden because they
        tend to attract either to each other or other taxa - and
        could in fact be on the shorter branches in the shortest
        tree (i.e., they are cryptic).  It only takes
        one long edge taxon to spoil the parsimony tree.  Cladistic
        parsimony is hobbled by their unmitigated influence in specific
        cases, and needs a leg up to avoid these and other perils.
        No philosophical justification of cladistic parsimony
        will ever change the informativeness of a single data set -
        that is predetermined by whatever processes of evolution have
        already occured to the characters, to the biologcial prowess
        of the investiagtor, and to a large extent by
        taxon sampling.  It's not only a question of the ability
        of the researcher to identify homology  - there are other
        factors that can lead to problems in the data.  But
        the clues to the problems in the data might notbe found
        in parsimony trees because they can reflect noise as
        signal, and signal as noise.

        More to the point of the thread, I consider it positivistic
        (read: like positivism) to presume that parsimony (or any other
        method of inference) will lead asymptotically to the truth when it
        has been known for so long that it can be positively mislead under
        some circumstances, and that adding more data will only
        make things worse.  To continue the myth of the sufficiency of
        parsimony is to say that it cannot be improved.

        James Lyons-Weiler