Dynamic Homology and Circularity

ROFFÉ, ARIEL JONATHAN

Oslo

Congreso; ISHPSSB 2019; 2019

International Society for the History, Philosophy, and Social Studies of Biology

A classical problem in the philosophy of systematics isthe alleged circularity in the determination of homologies. It is held that notevery trait is useful or informative for recognizing phylogeneticrelationships, but only traits that are homologous are. Thus, homologies haveto be identified prior to the beginning of phylogenetic analysis itself (i.e.detection of homologies is implicit in the construction of the data matrix thatfunctions as input to phylogenetic analysis). On the other hand, homologies areusually defined as traits that are derived from a trait in a common ancestor,which implies that phylogenetic relationships (and thus phylogenetic analysis)have to be established before homology recognition can take place. An also classical(by now) solution to this problem consists in distinguishing between twoconcepts of homology. Primary / topographical / hypothetical homologies aretraits that are similar in their structural (topographical or compositional)features, and are recognized prior to phylogenetic analysis by using theclassical Owenian criteria (or by multiple sequence alignment for moleculartraits). Secondary / phylogenetic homologies or homogenies are traits that arederived from a trait in a common ancestor and are recognized via phylogeneticanalysis. These analyses will typically reveal that not every primary homologyis a secondary homology, since they usually result in some convergence amongstructurally similar traits being present (i.e. primary and secondary homologyconcepts tend to have different extensions).Beginning in themid-1990s, a new methodological framework for the detection of homologies hasemerged, called the ?dynamic homology? approach. It initially came about formolecular traits (although it has now been extended to morphological traits aswell). The novelty consists in the fact that sequence alignment takes placeduring tree search and not prior to it, resulting in different alignments fordifferent trees. This blurs the line between primary and secondary homologyconcepts, and proponents have indeed suggested that we should get rid of thisdistinction. Although thesetechniques have potentially groundbreaking consequences for the circularitydebate and its standard solution, their philosophical significance for thisdebate has been mostly overlooked (especially by philosophers). Thispresentation is an attempt to examine the consequences of the dynamic homologyapproach for the circularity debate. I will hold two theses. First, that evenif the proponents were right and the distinction between primary and secondaryhomologies had to be discarded, there would be no reason to worry aboutcircularity within their approach. And second, that the idea that thisdistinction makes no sense is incorrect, because dynamic analyses still requirea data matrix that is generated by applying criteria that are prior andexternal to the analyses themselves.