The world’s a puzzle; no need to make sense out of it.
According to the philosophy of science by Thomas Kuhn, two more or less clearly defined phases are distinguished in the development of a scientific discipline [Kuhn 1962]. One of them is “revolutionary”: it involves a paradigm shift, which is a fundamental change of one exploratory model to another with respective change of the entire cognitive situation. Another phase is “normal”: it is associated with refining and solution of specific research tasks within the framework of the same paradigm with only a minimal change in the cognitive situation. These tasks are known as scientific puzzles [Gir'M 1973].
The taxonomic theories (TTs) functioning in biological systematics can be considered its paradigms; from a Kuhnian perspective, their development involves posing specific taxonomic puzzles and elaborating the means to solve them [Pavlinov 2019]. For instance, as far as general TT is concerned, the main conundrum is about understanding the Natural System: the various proposed ways of solving it led to scientific revolutions in systematics (see Chapter 2). For the aspect-based partial TTs embodying these different understandings, the respective puzzles are shaped by elaborating methodologies that are most consistent with certain understandings of the Natural System (typological, phylogenetic, etc.). For object-based and relational partial TTs, their puzzles are shaped by searching for answers to the questions of how to interpret species or homology, how to configure hierarchies, how to correlate kinship and similarity, etc.
In considering taxonomic puzzles, one of their important features should be pointed out. Unlike the “Kuhnian” ones, their main concern is the solution of specific tasks of a type similar to “find I don’t know what,” so they are problem puzzles. For example, it is clear that the classification sought should be natural, but it is not clear how to define it precisely and unambiguously: after all, we can only guess what the System of Nature is. Suppose we believe that such a classification should be phylogenetic, but it is not clear what precisely phylogeny is: we cannot observe it, so it remains only to assume some of its essential properties based on some other theoretical considerations and/or empirical observations. Thus, when considering the cognitive situation of systematics from this perspective, we have a kind of “taskbook” with a set of “tasks,” each with many unknowns, that make up the respective taxonomic puzzles. Moreover, it is completed with a kind of “solutions textbook” in the form of certain TTs—but there is nothing like definite “answers” at its end, with which one could check the definitive solutions to the puzzles. Instead, a spectrum of many possible “answers” appears during the attempt to solve the taxonomic puzzles.
This chapter explores some of the taxonomic puzzles associated with the basic concepts of systematics. Among them are problematic understandings of what the natural classification is and how it is structured; how taxon and character are interconnected by an iterative classification procedure; how similarity and kinship relate to each other; and how archetype, homology, and species can be treated.
Between Natural and Artificial Classifications
According to the basics of the general TT set out in Chapter 4, the main studied object of systematics is taxonomic diversity (TD) as a specific taxonomic Umwelt, and the main task is to describe this object in the form of the taxonomic system. The latter is thought of as an adequate cognitive model of TD, whereas in traditional (Linnaean) terminology it is usually called natural classification. In classical systematics, the latter is thought of as “globally natural” by reflecting the TD either as a whole or in its presumably most essential manifestation. With this, the conceptual history of systematics provided various particular understandings of this general notion, so its non-classical version presumes that different aspectual manifestations of TD are best represented by the corresponding “locally natural” particular classifications.
In any cognitive situation centered around the general conception of natural classification, regardless of its understanding, it is acknowledged that an exhaustive description of TD is theoretically unattainable. In fact, systematics develops particular classifications as more or less close approximations of the natural one; the least close, in contrast to the latter, are traditionally called artificial classifications. In general, all particular classifications are to some extent natural and to some extent artificial; the difference between them is quantitative, so the TD structure, however it is understood, may be represented by a certain set of classifications with different degrees of naturalness.
The main problem related to this global task of systematics is caused by the impossibility of gaining an unequivocal and commonly adopted understanding of what the natural classification is and how it can and should be attended. Because of this, the most fundamental and most problematic naturalness puzzle appears in systematics: it is caused, at a theoretical level, by a multiplicity of ways to understand the Natural System, and at a practical level, by a multiplicity of working classifications with various degrees of their “local naturalness.” It can be argued, without great exaggeration, that the conceptual history of systematics was driven by a search for ways to pose and solve this general puzzle. Obviously, these ways can be different depending on how specific cognitive situations are initially shaped onto-epistemically. Consideration of all questions related to this puzzle is within the competence of the TT in both general and particular understanding.
Some partial TTs refuse to consider this puzzle as loaded with a rich ontology. They develop an idea of general-purpose classifications interpreted quite pragmatically. Such classifications are the better, the wider the set of scientific or applied tasks they solve effectively [Gilmour 1940; Sokal and Sneath 1963]. However, they are sometimes called “Gilmour-natural.”
In classical natural science, the natural classification is likened to the law of nature [Duhem 1954; Rozova 1986; Zarenkov 1988; Zabrodin 1989; Vityaev and Kostin 1992; Rozov 1995; Mahner and Bunge 1997; Wilkins and Ebach 2014]. Such understanding is evidently inherited from the natural philosophy of the 17th and 18th centuries with its dominating idea of the law-like System of Nature; however, there is no universal understanding of what this “taxonomic law” might be. In one of its versions ascending to the scholastic tradition, classifications based on the essences are considered natural [Rozova 1986; Shatalkin 2012]. Close to this is a version according to which classification is natural if its taxa can be interpreted as natural kinds in the sense of Quine [Spencer 2016]. In phylogenetics, classification is natural if it represents phylogenetic pattern most adequately. In natural systematics of botanists, classification is natural if it is embodies many (essential) characters.
A generalization over different ways of solving this puzzle is offered by a rationalist approach, according to which it is necessary to elaborate and apply some explicit criteria of naturalness, or rather a system of coherent criteria of naturalness. Accordingly, it is presumed that a classification is more natural if it meets the conditions of a given criterion (system of criteria) of naturalness [Michener 1957; Rozova 1986; Pavlinov 2006, 2010b, 2018]. Each of them serves as a certain “calculus” for assessing the “local naturalness” of particular classifications. It presumes a certain “naturalness scale,” relative to which working classifications can be ordered from the most to the least natural.
Each particular TT (PTT) elaborates its own criterion (system of criteria) as part of its quasi-axiomatics based on the respective ontic and/or epistemic premises. In the case of ontology-based PTTs, particular Umwelts (particular TD manifestations) are first defined and then the particular systems of naturalness criteria of classifications are developed with respect to just these Umwelts. In contrast, PTTs with epistemic bases infer their criteria from certain methodological requirements, such as logical rigor. Thus, all systems of criteria of naturalness of classifications are theory-dependent and therefore local as attributes of certain PTTs. Therefore, any conclusions about the naturalness of classifications based on such criteria are also theory-dependent and local: a classification considered natural in the context of one PTT may appear artificial in other PTTs.
This conclusion seems to be relevant to an understanding of the significance of the criterion of predictivity of classifications as the fundamental criterion of their naturalness. It is usually considered of paramount importance by the taxonomists adherent of the physicalist philosophy of science [Gilmour 1940, Crowson 1970, Lyubishchev 1972; Rozova 1986; Starobogatov 1989; Marradi 1990; Rozov 1995], but the preceding consideration indicates it is no less “local” than any others. This is because the possibilities of reliable predictions (extrapolations) inferred from the respective classifications are obviously limited to the specific Umwelts they reflect [Pavlinov 2018]. This conclusion is compatible with the treatment of taxon as natural kind (in the sense of [Quine 1996]).
One of the general approaches to the solution to this puzzle elaborated by the rational cognitive program is based on the idea of a natural method: according to it, for a classification to be natural, it must be elaborated by means of such a method. Thus, the emphasis is shifted from the criteria of the naturalness of the classification itself to the criteria of naturalness of the classification method. With this, as far as ontology-based PTTs are concerned, the principle of methodological correspondence comes into effect; accordingly, the naturalness of a method is determined not on its own but with reference to the respective Umwelt. Therefore, in the search for a solution to this general puzzle, one must proceed from the comprehension of Nature (what it “really is”) to the comprehension of the respective natural method. It should be remembered that the most significant paradigmatic shifts in the conceptual history of systematics in the 18th and 19th centuries involved different treatments of the natural method following different treatments of Nature (see Sections 2.4 and 2.5).
Any system of criteria of naturalness of systematic methods includes an appropriate selection (“weighting”) of the classifying characters and the classification algorithms; in both cases, it depends by and large on the basic onto-epistemology. With respect to the characters, the most significant are those that most consistently indicate either certain features or relations of organisms considered essential within a certain PTT. Thus, recalling classical terminology, it can be said that one of the main tasks of the natural method is to elaborate the criteria for the selection of proper characters. As to classification algorithms, their “naturalness” is determined by their effectiveness as a means of elaborating classifications considered the most natural according to certain criteria.
These considerations can be demonstrably illustrated by two examples. If an Umwelt is based on the natural-philosophical idea of the overall interconnectedness of the elements of Nature, the respective natural method should be based on the consideration of “all and any” properties of organisms. Accordingly, the natural classification is elaborated either based on all available characters to produce a kind of “omnispective” one [Blackwelder 1967] or on the criterion of the compatibility of the characters (this goes back to the natural method of Adanson; see Section 2.4.2), according to which natural classification should be substantiated by the largest possible number of mutually compatible characters [Estabrook 1972; Le Quesne 1982]. In contrast, if an Umwelt is defined as the phylogenetic pattern, most attention should be paid to the correct selection of characters as most reliable indicators of monophyly (morphological or molecular), and to the elaboration of classification algorithms (say, parsimony or likelihood) to infer most effectively the “best” (for the given dataset) phylogenetic classification.
Facing a seemingly irreducible variety of “locally natural” classifications, the following important question arises: how' might it be possible to move from a set of local solutions to a global one? One possible answer to this question, and that claims to be a fairly general solution to the naturalness puzzle, may be development of the so-called faceted classification system, which is organized similarly to relational databases [Broughton 2006; Chan and Salaba 2016]. It allows different locally natural classifications to be combined into a single pool with the help of a certain meta-language that makes it possible to implement the principle of mutual interpretability’ of classifications. As a result, instead of searching for either a single, quite fuzzy “omnispective” classification that claims to reflect “all in one” or a certain “locally natural” classification proclaimed to be most significant according to a certain criterion (phenetic, phylogenetic, typological, etc.), we focus on principles of developing a faceted classification system that embraces all manifestations of TD and thus is capable of claiming to be considered “globally natural” [Pavlinov 2018, 2020].
-  Cited after Dan Milliman’s "Way of the Peaceful Warrior.”