Traits and/or states

pierre deleporte pierre.deleporte at UNIV-RENNES1.FR
Fri Oct 31 11:14:57 CST 2003

So there are two points at stake, why not... I don't see simple questions 
as stupid, neither do I see philosophical concerns as pedantic... at all... 
the trick with apparently basic questions is that they frequently require a 
full-fledged philosophical answer...

As for the different uses of "character, attribute, trait...", it seems to 
me that there is no universal acception for these terms. A dictionary of 
systematics should expose the different possible acceptions in different 
schools of systematics and by different authors. These acceptions are 
frequently linked to the way "features" of organisms (to use a possibly 
more general term) are used in the classificatory or in the phylogeny 
inference approach.
e.g., Jardine and Simpson, and after them Mickevich, used "attributes" for 
features not included in the data matrix for diverse reasons. Others will 
use "attributes" as features of organisms in a general acception...
Characters are frequently intended to mean features placed in the data 
matrix, features considered relevant for classification or for phylogeny 
Traits are sometimes used in a larger acception, like I am presently using 

This for the "current usage" point, and I'm not pretending to exhaustivity.

Now, for the philosophical concern of possible "differences of nature" (not 
simply differences of use) of "observed features".

Kirk suggests that there is no such difference between features observed on 
organisms, and goes to the point of considering that a column in the data 
matrix is NOT composed of different states of a same character...
Maybe we can see all these terms as equivalent at some level of generality. 
These are "observations" of aspects of reality, and thus interpretations of 
this reality in some more or less elaborated theoretical framework, as 
suggested by Kirk. This is of general value for our description of the 
world in a given theoretical context, and has nothing to do per se with a 
non-realistic vision of the world. Realistic theories intend to describe 
the existing real world.

But Kirk goes to the point of suggesting that a column in a data matrix is 
NOT a character and its different states. Maybe I don't understand his 
point, but it is puzzling for me, because the current acception, implicit 
or explicit, of a column of a data matrix is exactly this : a character and 
its different states. In other terms, different "avatars" of a same thing.
Unless we would logically not make a single column, but different ones, for 
every "state" (which some "structural classifiers" effectively suggest).

The explanation of this must come from the problematic, the theoretical 
context of the analysis. Just to consider phylogeny inference, it is based 
on a model of character evolution (Fitzhugh would likely call it the 
"explanatory law" of this phylogenetic analysis), according to which one is 
supposed to identify different occurences of a same, inherited character, 
beyond its different aspects. The optimality criterion for phylogeny 
inference (generally : optimizing homology = contiguity of identical 
character states on the cladogram) is applied to the different states of a 
same character, i.e. to the similarly, or differently, coded cells of a 
same column in the data matrix.
See also the notion of "hierarchy" in the analysis, in the last post of 
Zdenek Skala, which points the same way.

Is there any problem, Kirk ?


A 10:29 30/10/2003 -0800, Kirk Fitzhugh wrote:
>At 09:05 AM 10/30/03 -0600, you wrote:
>>At 04:36 PM 10/30/03 +0200, Matt Buys wrote:
>>>Is there a difference between states and traits and if so, what would
>>>that be?
>>A "trait" (character) is a feature of an organism in general, e.g., flower
>>color.  A state is the specific expression of that character in a given
>>individual, e.g., red, yellow, or white.
> From an epistemological standpoint, the distinctions between "character"
>and "state," or "states" and "traits" are not correct. All objects are
>perceived by way of their properties. To conceptualize or communicate
>perceptions, we infer subject-predicate relations, such as "The petals of
>this flower are red." But, there is no generic "character" such as "petal
>color" to which we assign a "state." It is senseless (no pun intended) to
>demarcate a "character" as something separate from what is called a "state"
>of that "character." If I say I observe a flower with red petals, then I
>have mentioned, in the form of a predicate, a property/character/trait,
>etc., of petals, which is the subject. In the parlance of cladistics, for
>instance, a data matrix is not composed of columns as "characters" and
>cells as "states." Rather, each cell summarizes observed subject-predicate
>relations among a group of individuals; the columns simply indicate the
>different subjects for which these relations have been observed.
>I have a slide show at the url listed below which discusses character coding.
>J. Kirk Fitzhugh, Ph.D.
>Associate Curator of Polychaetes
>Invertebrate Zoology Section
>Research & Collections Branch
>Los Angeles County Museum of Natural History
>900 Exposition Blvd
>Los Angeles CA 90007
>Phone:   213-763-3233
>FAX:     213-746-2999
>e-mail:  kfitzhug at

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

More information about the Taxacom mailing list