Fwd: Re: [TAXACOM] Characters.

[pierre deleporte]

A 23:29 20/06/2004 -1100, John Grehan wrote

>>At 08:42 PM 6/18/04 +0200, pierre deleporte wrote:
>>We will just have to agree to disagree. I view phenetic characters as 
>>characters that individually represent overall similarity of a feature. 
>>To me that's phenetic.

This acception is a Grehanian exception, and your persistent use of such a 
personal (unique, indeed) acception of the term impedes the discussion.
Anyway, "overall similarity" of a feature is of no use for distinguishing 
analyses: phenetic, cladistic, or else. See Richard Jensens's data matrix 
and explanations.
Any character or character state is an apriori hypothesis of "overall 
similarity", that is simply similarity, of a feature. The term "overall" 
makes no particular sense for a character state. How could this statement 
of similarity be "partial" ? Mystery...
This is why "phenetic = overall similarity" is applied by everybody (but 
you) to taxa, not to single character states.

"Overal" makes sense when you compare series of features in different 
objects, not when you compare a single feature. This I call 'logic' and 
useful acception of terms, and this is NOT contextual (see below).

>>The way I view it is that thin enamel is the outgroup condition - whether 
>>stated as absence of thick enamel or not, and thick enamel is apomorphic 
>>within the ingroup being analyzed so I would include thick enamel 
>>(whether or not one calls that a character state or a character - I don't 
>>really lose sleep on that one) in the character matrix.

The crucial point is not the name you give it, but the evolutionary model 
than one feature can evolutionarily change into the other feature: enamel 
can change from thin to thick, or the reverse.
The other crucial point is that the program will not support clades by 
symplesiomorphies, even if you put them in the data matrix, because the 
progeram is cladistic, not phenetic.

>>I don't ignore anything. All I am pointing out in all of this is that if 
>>you use characters that are not really apomorphies all the algorithmic 
>>contortions in the world cannot make a cladistic result out of 
>>non-cladistic data. You seem to have trouble understanding that.

By this very statement you prove that you are ignorant of the fact hat the 
algorithms do not make use of non-cladistic information in the data, even 
if you put "everything" into the data matrix. It is not because you give 
the complete data matrix thatthe program will make use of all of it without 
discrimination. Programs do not perform "contortions", they implement the 
cladistic logic. They have been devised by cladists for performing 
cladistic analysis. Not phenetic analysis. You obviously ignore that (and 
apparently don't want to learn).

>>>  if thick enamel were not an apomorphy for the great apes I would not 
>>> use it in analyzing human relationships.

Neither would the program, despite the fact that you put both states in the 
data matrix, because the program will root on the outgroup state, and thus 
use only the apomorphic
condition for supporting a clade, hence the program applies your very 
cladistic logic, so where is the problem?

>>Your logic and facts are not pure, they are contextual. That's pure 
>>Derrida perhaps.

OK this si pure post-modernism: if logics is contextual, and if the fact 
that cladistic programs are not phenetic is contextual, then you win. Note 
that you will always win. You can indefinitely play the game of developing 
a pseudo-discusssion, and always pretend that you're applying "your own 
contextual logic and reality of facts" (not only your own peculiar 
definitions, which are a great burden in the discussion, but can be 
overcome with some good will). And if so, science is dead, as Richard 
Jensen stated

But fortunately, logic is not contextual. Progress has been made between 
Aristotle and modern mathematics, but logics is logics and scientists play 
the game by the sames rules in this respect.

Not post-modern people of course: they like taking postures in different, 
pretendedly incommensurable contexts.

And cladistic programs perform cladistic analysis anyway you take them.
But a strict post-modernist may not care at all and pretend that logic is 
plural and incommensurable and that the fact does not exist in his own context.

>>>the cladistic programs do not make use of autapomorphies, contrary to 
>>>your wrong assertion. Will you deny it, and prove your point?
>>
>>I wasn't saying that they do!

Of course you did and you persist, you said that if autapomorphies are put 
into the data matrix, the "contorsions" of the program will not perform a 
cladistic analysis. You stated this and this is false, as logically 
demonstrated on Richard's example.

>>Because one would not be using the same data set. To do a cladistic 
>>analysis I would use a different data set from that proposed in other 
>>supposedly cladistic analyses because those analyses used characters that 
>>are not apomorphies (they were treated as if they were apomorphies, but 
>>they were not).

You are simply wrong. The program does not make use of non-cladistic 
information. The program "reduces" by itself the complete data set to the 
cladistically informative one. You are ignorant of what the programs do.

>>But I would not include plesiomorphies.

Don't "include" putatively plesiomorphic STATES if you like, but you'll get 
the same result as the program, which makes no "apomorphic" use of the 
plesiomorphies you have introduced because it implements the outgroup 
criterion exactly as you would do by hand.

>>The garbage I am referring to is the use of characters that are being 
>>accepted as representing apomorphic states when they are not.

 From the moment when you put the outgropup in the anaysis, nothing of this 
kind occurs. Now you can debate on which outgroups should be used, but this 
has nothing to do with your persisting assertion that cladistic programs 
perform phenetic analysis (and by the way you throw away all molecular 
phylogenies, which is not a minor point).
Here you come back to the case of primates, when the discussion is on your 
rejection of molecular data on the wrong basis that programs would be 
phenetic because the data matrix is complete. Which is a complete grehanian 
mistake, apparently by sheer ignorance of the cladistic algorithms.

>>of course postmodern absolute relativism is another planet, and on this 
>>planet you get different logics and different sciences superbly ignoring 
>>one another, and everybody is happy in his own peculiar, subjective "real 
>>world".
>Don't understand any of this.

Thus I translate: your lasting way of using not only highly peculiar 
definitions (e.g. "phenetic" for qualifying character states instead of 
taxa) but also pretending that logic and facts are contextual is typically 
"post-modern". It impedes any rational discussion between people pretending 
to stand in different "contexts" while dealing with the same topic, and if 
this attitude is popularized it leads to the end of science and rational 
debate. Because you'll get several "sciences" and "rationalities" for a 
same topic (here, phylogeny
inference) with no possible communication and progress toward common 
understanding (see more on Richard's web page).
This is why present post-modernism is the shame of French philosophy (I'm 
French, no offense), the more in the historical land of the enlightenments 
which promoted rationality, materialism and universalism, not "contextual 
logic" promoted by some fashioned text anaysists.
Now it's not me who quoted Derrida, and thus appealed to postmodernism for 
talking of science, and pretended that logic is contextual: it's you.
D'Hollbach preserves, I'm hopefully pre-postmodern...

Pierre


Pierre Deleporte
CNRS UMR 6552 - Station Biologique de Paimpont
F-35380 Paimpont   FRANCE
Téléphone : 02 99 61 81 66
Télécopie : 02 99 61 81 88