The phylogenetic literature that deals with species, taxa, as well as with homologs, is replete with dismissals of taxa, or homologs, as natural kinds, for these are considered abstract entities, like sets, or classes (Hull 1999). The same is true of properties (attributes), which are dismissed as universals that cannot be instantiated by evolutionary entities (Grant and Kluge 2004). Hey (2001 p 153) for example dismisses property cluster kinds as polythetic classes, where "a polythetic class is just a category in which the members need only meet some, not all, of the membership criteria.'' But as emphasized previously: for property cluster kinds (as well as for other kinds) to be natural requires that their properties are relevant to causal processes, i.e., that tokens of kinds take part in causal processes by virtue of their properties (Sober 1981; Boyd 1999; Shoemaker 2003; LaPorte 2004; Lipton 2004).
Homologs (Grant and Kluge 2004), as well as species (Hull 1999), have been considered to be particulars (individuals). This move was motivated by a certain reading of Hennig (1950, 1966), which found him to have replaced "intensional'' with "extensional'' thinking in systematics (Dupuis 1984). Such reading of Hennig has led to the assertion that systematics must rely on "ostension'' and "extensionality'' (Kluge 2003a). The meaning of extensionality as used in this context remains obscure (Dupuis 1984 p 13: "Hennigian taxonomy is extensional since, for appraisal of each attribute, it requires an extensive consideration of various objects''), whereas the relation of ostension is more easily understandable. In the simplest case, ostension is the relation of "pointing at'' an object in order to establish that object as the referent of a proper name, or of a sentence of the subject-prediacte structure (Fx). But for Grant and Kluge (2004), it is not objects and their properties that serve as homologs. Instead, "only the transformation from one character state to another, a ! a0, constitutes evidence in phylogenetic inference'' (Grant and Kluge 2004 p 25). If only transformation events constitute evidence, and if systematics is to be ostensional, then systematists would have to be able to "point at'' such transformation events.
The collapse of a bridge or the burning of a house are events that witnesses can point at, but the transformation events that are of interest to systematists are events of the past. What is required, therefore, is identification of character states and inference of their transformation. "However, character state transformation has always been treated primarily as a metaphor Furthermore, character states themselves can be seen as manifestations of the transformational metaphor'' (Weston 2000 p 135). In this quote, I would replace the term "metaphor'' with that of "concept,'' for as convincingly argued by Hanson (1961), perception itself entails a conceptual component (Rieppel 2004a). As noted earlier, the mere pointing at a rabbit and the utterance of "gavagai'' by a native speaker does not tell the English speaking audience what the native speaker means by gavagai: a rabbit, an undetached rabbit-part, or a rodent (Quine 1964)? It seems impossible to meaningfully talk about the world without concepts (Luntley 1999).
A transformation series of character states purported to be homologous does not result from mere ostension but from character conceptualization ("transformation series analysis ... begins with the construction of an initial transformation series'': Kitching et al. 1998 p 218; my emphasis). Hennig (1966 p 94) was very clear on this issue: "the belonging of characters to a phylogenetic transformation series ... cannot be directly determined.'' To solve the problem, Hennig (1966 p 94) turned to an account (in Cain and Harrison 1958 pp 86-88) of J.H. Woodger's analysis of morphological comparison in the language of logical positivism (Rieppel 2003b), which "to that extent is exact'' (Hennig 1966 p 94). According to Woodger (Cain and Harrison 1958), morphological comparison individuates parts of organisms in terms of their relation, and compares them in terms of their properties. The description of properties of, and relations between, objects are the cornerstones of the positivist philosopher Rudolf Carnap's "construction theory,'' i.e., the theory of how science reconstructs the logical structure of the world (Mayhall 2002). As argued earlier, the same theoretical background underlies Hennig's (1957, 1966) conceptualization of the cladogram (Rieppel 2003b). There is no need to look back on Carnap, Woodger, or even Hennig, however, to understand that the identification of characters, character states, and transformation series entails a conceptual component (Hanson's (1961) work was, after all, a stab at the positivists' adherence to something akin to "simple'' observation). In his talk announcing his own transformation, Farris (Mitter 1980 p 188; see also Dupuis 1984 p 16) asserted that "morphologists do not sample characters, they synthesize them.'' If "extensional'' systematics must be based on ostension, but if it admits only character transformation as evidence, then the argument boils down to the claim that systematists must be able to "point at''
concepts. This conclusion merely shows that such arguments confound concepts with objects (Mahner and Bunge 1997).
The discussion of transformation series reflects back on the "transformational" versus "taxic" approach to homology that was hinted at by West-Eberhard (2003) in her distinction of broad-sense versus cladistic homology. The essence of this debate (that started with Patterson 1982) is that the transformational approach allows homology to encompass both symplesiomorphy and synapo-morphy, whereas the taxic approach requires homology to be synonymous with synapomorphy (it requires a statement of homology to be coextensive with a statement of synapomorphy). Wiesemuller et al. (2003 p 89) consider the taxic approach as too coarse grained because it blurs the distinction of plesiomorphy and apomorphy, in addition to the fact that the concept of homology cannot easily be applied to "absence"-characters. The latter problem was (arguably) "solved" semantically: snakes do not have "no legs,'' instead they have "modified legs,'' i.e., "lost legs'' (Platnick 1978; see the discussion of the deduction of different meanings from different premises in Rieppel 2004b p 79). The distinction of homology from symplesiomorphy and synapomorphy on the other hand is related to the qua-problem. "Bird wings and bat wings are homologous'' violates the requirement for coextensionality of statements of homology and of synapomorphy because it implies a symplesiomorphy: birds and bats share tetrapod (fore)limbs, not wings. Patterson's (1982) goal simply was to remove ambiguity, and render statements of homology testable by congruence, in claiming that homology statements should always be assessed as (qua) statements of synapomorphy—since what is a symplesiomorphy at one level (jaws of amniotes) becomes a synapomorphy at a more inclusive level (jaws of gnathostomes). Interesting evolutionary talk about a comparison of bird wings with bat wings would be talk about convergence, but then bird wings and bat wings could not be homologous.
Was this article helpful?