Stud. Hist. Phil. Biol. & Biomed. Sci.

37 (2006) 771791

Studies in History and Philosophy of Biological and Biomedical Sciences

www.elsevier.com/locate/shpsc

Organisms as natural purposes: The contemporary evolutionary perspective

D.M. Walsh
Department of Philosophy and Institute for the History and Philosophy of Science and Technology, University of Toronto, 91 Charles Street West, Toronto M5S 1K7, Canada

Abstract Kants conception of organisms as natural purposes raises a challenge to the adequacy of mechanistic explanation in biology. Certain features of organisms appear to be inexplicable by appeal to mechanical law alone. Some biological phenomena, it seems, can only be accounted for teleologically. Contemporary evolutionary biology has by and large ignored this challenge. It is widely held that Darwins theory of natural selection gives us an adequate, wholly mechanical account of the nature of organisms. In contemporary biology, the category of the organism plays virtually no explanatory role. Contemporary evolutionary biology is a science of sub-organismal entitiesreplicators. I argue that recent advances in developmental biology demonstrate the inadequacy of suborganismal mechanism. The category of the organism, construed as a natural purpose should play an ineliminable role in explaining ontogenetic development and adaptive evolution. According to Kant the natural purposiveness of organisms cannot be demonstrated to be an objective principle in nature, nor can purposiveness gure in genuine explain. I attempt to argue, by appeal to recent work on self-organization, that the purposiveness of organisms is a natural phenomenon, and, by appeal to the apparatus of invariance explanation, that biological purposiveness provides genuine, ineliminable biological explanations. 2006 Elsevier Ltd. All rights reserved.
Keywords: Kant; Purpose; Teleology; Mechanism; Self-organization; Replicator biology; Development; Invariance; Explanation

E-mail address: denis.walsh@utoronto.ca 1369-8486/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.shpsc.2006.09.009

772

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

0. Introduction Kants third critique, in particular the Analytic of the teleological power of judgment, famously introduces a tension between our conception organisms and our understanding of the natural world as governed by unied mechanical law. Organisms, Kant tells us, are natural purposes and as such are subject to teleological explanation. We account for the natures and capacities of their constituent parts and processes by appeal to those purposes of the organism that they fulll. At the same time, organisms are natural entities subject to mechanical laws. Mechanical laws give us complete scientic explanations of all the phenomena of the world and concedes no irreducible explanatory role to goals or purposes. The tension gives rise to Kants famous Antinomy of the teleological power of judgmentroughly, that organisms both must be and cannot be judged to be wholly the products of mechanical processes. The tension Kant articulates regarding organisms as natural purposes is as germane today as it ever was. Organisms really do have the astonishing properties by which Kant identies them as natural purposes. Moreover, scientic explanation, today just as in Kants time, is cast in the mechanical mode. The tension ought to survive in contemporary biology, perhaps not as an antinomy, but nevertheless as a genuine challenge to the adequacy of our conception of the biological world. My objective here is to draw to the attention of contemporary biology (and philosophy of biology) the relevance of Kants conception of organisms as natural purposes. Kants problematic may have been largely forgotten by contemporary biology, but it has strong resonances with issues that are only now beginning to attract biologists attentionsselforganization, the emergent properties of organisms, their adaptability, their capacity to regulate their component parts and processes. I do not attempt to persuade biologists to adopt the critical philosophy. Indeed there is a question whether it is consistent with modern day naturalism.1 I do claim that Kants Critique of the teleological power of judgment oers a particularly vivid articulation of the challenge raised to any theoretical biology by the nature of organisms, a challenge encapsulated in the Antinomy of the teleological power of judgment. I contend that contemporary biology has (inadvertently) embroiled itself in an analogous antinomy, one of teleological explanation. Recent research on organismal development suggests that biological phenomena cannot be wholly explained unless we consider organisms as both mechanical and purposive entities. I would like to suggest, however, that biologists have the conceptual resources, without recourse to the critical philosophy, to reconcile the mechanical nature of biological processes with the fact that the conception of organisms as natural purposes has an ineliminable explanatory role to play. 1. The Antinomy of the teleological power of judgment An organism, according to Kant, constitutes a functional unity whose parts and processes contribute to its development, sustenance and reproduction, indeed whose parts seem to exist in the form they do precisely because they full these roles. Organisms are distinguished by three diagnostic features: (i) self-organization: the growth and interaction of parts is evidently caused by the organizational structure of the organism as a whole and is directed
1

See John Zammitos discussion (2006).

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

773

toward the attainment of a viable adult form; (ii) self-reproduction: each organism is capable of securing the existence of other such organisms similar in structure and function; (iii) and self-nourishment (McLaughlin, 2000). During its ontogeny, Kant tells us, an organism rst processes the matter that it adds to itself into a specically-distinct quality, which the mechanism of nature outside of it cannot provide, and develops itself further by means of a material which, in its composition, is its own product. (Kant, 2000, p. 371)2 The constituent parts and processes of a living thing are related to the organism as a whole by a kind of reciprocal causation. The essential denition Kant oered of : : : organic form : : : was that of the reciprocal interrelation of parts as means to ends, and consequently, of the priority of the whole over the parts in the constitution of the entity. (Zammito, 1991, p. 218) The denition of an organic body is that it is a body, every part of which is there for the sake of the other (reciprocally as an end, and at the same time, means). : : : An organic (articulated) body is one in which each part, with its moving force, necessarily relates to the whole (to each part in its composition). (Opus postumum, quoted in Guyer, 2005, p. 104) I would provisionally say that a thing exists as a natural end if it is cause and eect of itself. (Kant, 2000, p. 371) The self-motivating, self-forming capacities of organisms, the reciprocal dependence of part on whole and whole on parts, the supple goal-directed, compensatory capacities characteristic of organismal development, these, Kant tells us are all essential to our concept of an organism to think of an organism qua organism is to think of it as a purposive entity. Moreover, an organisms purposes are immanent in it; they are not extrinsic purposes, like those of artifacts. Hence, organisms are natural purposes. Organized beings are thus the only ones in nature which : : : must nevertheless be thought of as possible only as ends, and which thus rst provide the objective reality for the concept of an end that is not a practical end but an end of nature, and thereby provide natural science with the basis for a teleology : : : (Ibid., pp. 375376) Kants concern about this conception of organisms is that while it is obligatorywe cannot take ourselves to be studying organisms qua organisms unless we consider them as natural purposesit seems to put the dening features of organisms beyond the ambit of scientic explanation. For Kant all genuine explanation is mechanical. To explain mechanically the properties of a naturally occurring entity, one cites the properties of its parts that determine the whole. His conception of scientic explanation appears to conform to what C. D. Broad has called Pure Mechanism, according to which, there is: : : : (a) a single kind of stu, : : : (b) a single fundamental kind of change, : : : (c) a single elementary causal law, : : : and (d) a single and simple principle of composition,
All citations from the Critique of judgement are taken from the translation by P. Guyer & E. Matthews, Kant (2000). References give the pages in the Akademie edition.
2

774

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

according to which the behaviour of any aggregate of particles, or the inuence of any one aggregate on any other, follows in a uniform way from the mutual inuences of the constituent particles taken by pairs. (Broad, 1925, p. 45) On the one hand, the functional unity of an organism can be explained in the mechanical mode; an organism is a causal consequence of the capacities of its parts. But, on the other hand, the parts themselves seem to be the eects of the structural and functional unity of the organism. Consequently, we cannot account for the nature of organisms by mere mechanical explanation alone: : : : it is quite certain that we can never adequately come to know the organized beings and their internal possibility in accordance with merely mechanical principles of nature, let alone explain them; and indeed this is so certain that we can boldly say that it would be absurd for humans even to make such an attempt or to hope that there may yet arise a Newton who could make comprehensible even the generation of the a blade of grass according to the natural laws that no intention has ordered; rather, we must absolutely deny this insight to human beings. (Kant, 2000, p. 400) This tension between the requirement to conceive of organisms as natural phenomena, explicable in the mechanical mode, and the requirement to think of them as purposive entities whose natures defy mechanical explanation, is articulated in The Antinomy of the teleological power of judgment: The rst maxim of the power of judgment is the thesis: All generation of material things and their forms must be judged as possible in accordance with merely mechanical laws. The second maxim is the antithesis: Some products of material nature cannot be judged as possible according to merely mechanical laws (judging them requires an entirely dierent law of causality, namely that of nal causes). (Ibid., p. 387). 2. Sub-organismal biology Contemporary biology has responded to the problem posed by the natural purposiveness of organisms by the simple expedient of ignoring it. There is a general thought nowadays that biology has no need of teleological thinking. There are a couple of reasons. The rst is that the project of contemporary comparative biology is substantially dierent from that of Kants time. The second is that the range of explanatory resources available to biologists has expanded signicantly. These reasons both derive from the inuence of Darwin. Darwins most signicant achievement was the understanding that biologys two great explanandathe t of organisms to their conditions of existence and the diversity of organic formare both the consequence of a single mechanical process, natural selection. That process is not to be found occurring within organisms, but within populations. Selection causes populations to change their constitution over time, which in turn realizes the increase in both the adaptedness and diversity of organisms. Darwins theory of selection changes the explanatory project of biology from that of explaining the natures of organisms by appeal to some principle immanent in organisms themselves to one of explaining the nature of organisms by appeal to a causal process occurring within populations. The

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

775

problematic features of organisms that impelled Kant to identify them as natural purposesthe exquisite integration of organisms, their adaptive plasticity, their self-organizing capacitiesare simple causal consequences of change in population structure wrought by natural selection. This line of thought led Michael Ghiselin to proclaim Darwins triumph over teleology: he developed a new way of thinking that allows us to dispense altogether with that metaphysical delusion (Ghiselin, 1993, p. 483) and occasioned Ernst Mayrs insistence that Darwin had solved Kants great puzzle (Mayr, 1988, p. 58). Darwins theory is explicitly structured according to the mechanistic model (Cornell, 1986; Hull, 2003). Darwin took Herschels Preliminary discourse (1987) as the canonical source on the proper conduct of scientic enquiry. Herschel, in turn, held Newtons mechanics and Lyells geology as paradigms of the scientic method. Herschel was particularly impressed with (what he took to be) the insistence by Newton and Lyell on vera causae explanations. A vera causa explanation, according to Herschel, invokes only those processes as causes that can be directly observed. Moreover, they must be observable independently of the eect to be explained. The degree of eect of a true cause must vary with the degree of the cause itself. Darwin prided himself on the conformance of his theory of natural selection to Herschels methodological precepts. Variation and struggle for existence are directly observable in a population, quite independently of their inuence on population change. They are sucient to cause population change. The rate of change in a population is proportional to the degree of variation in a population and the severity of the struggle. Thus, in Darwins entangled bank all its various organisms, : : : so dierent from each other, and so dependent on each other in so complex a manner, have all been produced by laws acting around us (Darwin, 1972, p. 462). Darwins theory is a strictly causal, mechanical one. Natural selection allows us to assert the thesis of the antinomy and prescind from the antithesis. It does so only if those features of organisms that so impressed Kant have no indispensable explanatory role to play in biology. As John Cornell expresses it: To reject Kants argument we would have to explain away the organism (Cornell, 1986, p. 408). This is precisely what much of contemporary biology has undertaken to do. Current evolutionary biology is largely directed toward the study of how supra-organismal entitiespopulationschange as a result of the causal powers of sub-organismal entitiesreplicators. The most thorough and provocative exponent of the sub-organismal nature of biology is Richard Dawkins: Evolution is the external and visible manifestation of the survival of alternative replicators : : : Genes are replicators; organisms : : : are best not regarded as replicators; they are vehicles in which replicators travel about. Replicator selection is the process by which some replicators survive at the expense of others. Vehicle selection is the process by which some vehicles are more successful than others in ensuring the survival of their replicators. (Dawkins, 1982, p. 82) Organisms are mere middlemen in evolution, a sort of interface between the organismbuilding activities of replicators and the selecting role of the environment. Their salient characteristics, and indeed their very existence, are to be explained by appeal to the causal capacities of replicators. The integrated multi-cellular organism is a phenomenon which has emerged as a result of natural selection on primitively independent selsh replicators. It has paid replicators to behave gregariously. (Ibid., p. 264)

776

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

There are good reasons why replicators, rather than organisms, enjoy this privileged position. They play a distinctive unifying role in our account of evolution. Evolution comprises three constituent processes: inheritance of traits, ontogeny of organisms and change within populations. Replicator biology oers us an elegant account of how these three processes taken together lead to the t and diversity of organisms. (i) Inheritance: Evolution requires variation in heritable traits. Replicators, we are told, are the units of inheritance. They are the only entities that are both copied within parents and transmitted from parent to ospring in reproduction. (ii) Ontogeny: Development, according to the sub-organismal view, is simply the implementation of a developmental program, the expression of phenotypic information encoded, or programmed, in the replicators (e.g. Maynard Smith, 2000). Organisms may be what are selected, but replicators build organisms. (iii) Population Change: Evolution by natural selection occurs when populations change their structure as a consequence of the dierential survival and reproduction of organism. But not just any old change in population structure counts as evolutionary change, only changes in the relative frequency of replicators. Indeed natural selection is usually dened as changes in replicator frequencies that result from the dierential contribution of replicators to survival and reproduction (Bell, 2000). Replicators, then, are the only entities that take part in all the constituent processes of evolution inheritance, ontogeny and population changeand whose activities unite these processes into the process of evolution. Replicator biology also furnishes a ready explanation of the adaptiveness of evolution. The adaptedness of organisms is the result of the accretion of adaptive (that is, selected) replicators. Fitness-promoting replicators are assembled by selection into integrated suites of traits (Ayala, 1970). The appearance of biological design is the consequence of an iterated process of the introduction of variant replicators, their recombination and selection. Here again, the inuences of Pure Mechanism run deep. Sub-organismal biologys proprietary processes and explanations are mechanical, one and all. The properties of organisms are explained by appeal to the capacities of their parts. The parts of organism (replicators) have an explanatory and ontological primacy over the whole. Replicators exist. That is fundamental. Phenotypic manifestations of them : : : may be expected to function as tools to keep replicators existing. Organisms are huge and complex assemblages of such tools : : : (Dawkins 1986, p. 263). The success of sub-organismal biology depends upon replicators having the sort of eects that in Newtons words should esteem them in the highest (Newton, 1963, p. 398) There are basically three: (i) Constancy of quality: The role of replicators requires that each replicator has a reasonably constant, or predictable, phenotypic eect across contexts. Replicators ground the inheritance of phenotypes only if their eects are reliably constant across dierent organisms, i.e. across associations with dierent replicators; (ii) Constancy of quantity: The tness consequences of a trait must be reasonably constant across contexts. Were this not the case evolution by natural selection could not be cumulative.3 (iii) Independence: The constancy of phenotypic eect across contexts requires that replicators exert their eects on phenotype more or less independently of one another.4
3 See Kauman (1995) on the conditions for populations of organisms to adapt on smooth, multiply-peaked landscapes. 4 Fisher (1930) makes the independence assumption explicitly.

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

777

The eects of individual replicators on an organisms phenotype are thus decomposable. The degree of similarity of organism is a function of the degree of similarity in their replicators. Given the enormous number of replicators in each organism, it more or less follows from (i)(iii) that the eect of each replicator on tness is on average small. So, the adverse eect of new mutant replicators will generally be minimal. Taken together these conditions tell us that organisms are, in Broads terminology, resultant eects of the capacities of their replicators. Replicators are vera causae of organisms (at least in Herschels sense).5 Two further conditions must be met by replicators if populations of them are to undergo adaptive evolution. Lewontin has dubbed these continuity and quasiindependence: By continuity we mean that very small changes in character result in very small changes in : : : reproductive tness. Neighborhoods in character space map into neighborhoods in tness space. : : : By quasi-independence we mean that there exist a large variety of paths by which a given character may change, and : : :, a non-neglible proportion of these paths will not result in countervailing eects of sucient magnitude to overcome the increase in tness from the adaptation. (Lewontin, 1978, p. 247) Populations of replicators certainly meet these conditions. Because replicators provide us with mechanical explanation for both the natures of organisms and the process of evolution, there should be no residual temptation to teleology (Cummins, 2002). Sub-organismal biology has its detractors. Most attempts to dislodge it from its privileged position in biology involve pointing out the degree to which the above assumptions are violated (for example, epistasis) or that the adaptation promoting power of replicator selection is curtailed (Amundson, 2001; Gould & Lewontin, 1979). Another strategy is to argue that replicator biologys claim to comprehensiveness is ill deserved. Replicators are not the only units of inheritance. Acquired morphological traits, behaviours, culturally transmitted capacities can all be inherited and selected without replication (Boyd & Richardson, 2004; Jablonka & Lamb, 2005; Mameli, 2004). Nor are sub-organismal replicators the sole determinants of phenotype; the control of development is distributed throughout the organism/environment system (Griths & Gray, 2001; Oyama, 1985; Oyama, Grifths, & Gray, 2001; Moss, 2002). These are important checks against what appears to have been an almost uncritical embrace of sub-organismal biology in the late 20th century. Nevertheless, sub-organismal replicators do play an important causal role in inheritance, even if they are not the only things that do. They often correlate strongly with both tness and phenotype. They are at least important dierence makers in the development of a wide range of phenotypes. So, wholly comprehensive or not, replicators have a central explanatory role to play in contemporary biology. The pressing question, one that contemporary biology has largely declined to broach, is what conditions secure the capacities of replicators to act in this way? A tempting answer is that replicators possess these remarkable capacities by their natures. A tradition inaugurated by Max Delbruck maintains that the startling stability of DNA and its capacity to make high-delity copies are inherent in the structure of the genetic material; these are what make life possible. A starkly dierent answer, one that seems just recently to

Peter McLaughlin has pointed out to me how far short replicators fall as genuine Newtonian vera causae.

778

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

be gathering some empirical support, is that replicators have their remarkable capacities because of the capacities of the organisms of which they are a part. 3. Organismal biology Against the tide of sub-organismal mechanism of the 20th century biology, a few biologists, for example Waddington (1957) and Schmalhausen (1949), continued to stress the signicance of features of entire organisms for our understanding of biological processes (Sarkar & Gilbert, 2000). The features of organisms as natural purposes were considered of particular signicance: you cannot even think of an organism : : : without taking into account what variously and rather loosely is called adaptiveness, purposiveness, goal seeking and the like (Von Bertalany, 1969, p. 45). Organicism is once again getting a hearing, at least among some biologists, due to the recent resurgence of interest in organismal development. The most explanatorily signicant feature of organisms is their plasticity. Plasticity, according to Mary Jane West-Eberhard (2005a, 2003) is a universal principle of all living things. It consists in : : : the ability of an organism to react to an internal or external environmental input with a change in form, state, movement, or rate of activity (ibid., p. 33). West-Eberhards account of plasticity is strongly reminiscent of Kants description of the distinctive features of organisms. Plasticity consists in the adaptive capacity of organisms to regulate their developmental sub-systems in order to build and maintain a stable, working organism. The plasticity of organisms is pivotal in adaptive evolution. Plasticity engenders robustness, the capacity to maintain viability by making compensatory, adaptive changes to phenotype. This in turn endows organisms with a broad phenotypic repertoire; an organism is capable of producing any of a wide range of stable phenotypic outcomes. According to West-Eberhard plasticity and the phenotypic repertoire of organisms are the principal causes of adaptive evolution. On this view, adaptive evolution proceeds in two stages phenotypic accommodation followed by genetic accommodationboth of which depend upon the plasticity of organisms. West-Eberhard denes phenotypic accommodation as the immediate adaptive adjustment of phenotype to the production of a novel trait or trait combination (ibid., p. 147). She denes genetic accommodation as adaptive evolution that involves gene-frequency change (West-Eberhard, 2005b, p. 6544). The process, as envisaged by West-Eberhard, goes as follows. A mutation or environmental change, or change in development produces a new phenotype through phenotypic accommodation. The organism responds to the new circumstances by the production of a novel phenotype. Given the phenotypic repertoire of organisms, there is latent genetic and developmental variation in a population for the production of any particular phenotype. The introduction of a novel phenotype exposes this variation. Some gene combinations or developmental structures may underwrite more robust, reliable or ecient ways of producing a novel adaptive phenotype. The eect is that the threshold for the development of the trait is lowered (Waddington, 1957). Recombination and selection of these developmental and genetic variants occurs. The consequence is evolutionchange in genetic structure of the populationas a consequence of the developmental plasticity of organisms. There are two important features that distinguish this conception adaptive evolution from the version oered by replicator biology. The rst is that it is a property of organismstheir plasticitythat explains why evolution is adaptive. The second is that novel

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

779

adaptive phenotypes are not caused by changes in replicator frequency, as the sub-organismal model has it. On the contrary, adaptive novelties, brought about by plasticity of organisms, cause changes replicator frequencies. Without developmental plasticity, the bare genes and the impositions of the environment would have no eect and no importance for evolution. (West-Eberhard, 2005b, p. 6544) 3.1. Modularity of development The key to the adaptive plasticity of organisms lies in the architecture of development: development is modular. Developmental modules have two dening features, their internal structure and their mutual dissociation. The integrated internal structure of modules makes them robust; they are capable of producing their characteristic output despite the vagaries of their contexts, and despite changes to their structure (von Dassow & Munro, 1999). Gene modules oer a vivid example. To the extent that genes exert control over phenotype it is not through individual genes working in isolation; genetic control of development is exerted through suites (modules) of genes working in concert. Genetic modules are characterised by complex regulatory feedback relations among the module components. One signicant consequence is that gene modules exhibit a signicant degree of robustness. The function of a module can withstand the loss or malfunction of one or more of its component genes. Consequently, each gene module produces its characteristic output across an enormous range of conditions (Gibson, 2002; von Dassow et al., 2000). This buering also turns gene networks into evolutionary capacitors (Bergman & Siegal, 2002). Gene networks are capable of producing novel stable products in novel circumstances (Greenspan, 2002). The dissociation of modulesthe nature of the regulatory relations among them exerts another kind of phenotypic eect. The repertoire of a developmental module, the range of its capacities taken in isolation, is considerably greater than the range of its realized eects. Typically each module is capable of producing any number of a large array of stable outputs (Von Dassow et al., 2000). Which of its capacities is manifested on a particular occasion is determined by the context in which the module nds itself. Modules are arranged in interacting hierarchies. Typically, each module directly inuences and is inuenced by other modules (but not many). The plasticity of development not only permits the adaptability of organisms, it also secures continuity, quasi-independence and high delity of replication, those conditions necessary for replicators to act as they are required to do by the precepts of our mechanistic replicator biology. Continuity is caused by the buering (or canalization) brought about by the modular structure of development causes continuity. It guarantees that a gene, or gene network, will have similar phenotypic eects across contexts, its broad phenotypic repertoire notwithstanding. The robustness of development produces phenotypes that are remarkably constant despite the enormous amount of underlying genetic variation within the population, and the unpredictability of environments. Furthermore, ontogenetic buering enables organisms to consolidate the developmental roles of genes; the actions of genes become routinised, locked into the production of particular phenotypes (Newman & Muller, 2001). Genes (replicators) may exert phenotypic control during development and ensure the reliable inheritance of phenotypes, but only because the entire

780

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

organism exerts regulative control over the activities of genes. As a consequence of the buering of development, genetic dierences between individuals are largely damped out. Each replicator, then, has a minor eect on overall organismal tness. The modular structure of development also promotes the quasi-independence of replicators. Those genes with discernible phenotypic eects will tend to be strongly dissociated from the eects of others because of the modularity of development (West-Eberhard, 2005b). Modularity quarantines one part of the phenotype against the deleterious eects of changes to their parts. Phenotypic accommodation reduces the amount of functional disruption occasioned by developmental novelty (West-Eberhard, 2003, p. 147). In securing the conditions necessary for the stability of organisms, the plasticity of development, also puts into place the requisite conditions for replicators, genes, to serve as units of inheritance, phenotypic control and as units whose change in relative frequencies is a mark of adaptive evolution. The marvellous capacities of genes may be not so much a pre-requisite for the evolution of organisms as a consequence of the evolution of organisms. The regulation of gene action by organisms seems also to be crucial in ensuring the high delity of replication. Gene replication is highly fallible, but replication errors are actively repaired by the regulative processes of the cell (Fox Keller, 2000, p. 32). Robert Haynes writes: The stability of genes now is seen to be more a matter of biochemical dynamics, than of molecular statics of DNA structure. The genetic machinery of the cell provide the most striking example known of a highly reliable, dynamic system built from vulnerable parts. (Haynes, 1988, p. 577) Wagner (2005) claims that the robustness of DNA is a highly evolved adaptation of organisms. Newman and Mueller surmise that : : : the correlation of an organisms form with its genotype, rather than being a dening condition of morphological evolution, is a highly derived property (Newman & Muller, 2001, p. 576). 3.2. Organismal and sub-organismal biology The organism-centred conception of evolutionary biology gives us a radically dierent perspective on the processes of development, inheritance and adaptive evolution from that aorded by sub-organismal biology. According to organism-centred biology it is an irreducible property of organisms as wholes that explains the processes of development, inheritance and adaptive evolution. This is not to insist that replicators do not play a role. Genes really are (among) the units of inheritance. Replicators (genes) really do exert phenotypic control (along with myriad other developmental resources). Evolution does consist, inter alia, in the change in gene frequencies in a population. On the organism-centred view, the capacities of organisms, by which Kant identies them as natural purposestheir self-organization, self-regulating, goal-directed capacitiesconstitute the ground for the possibility of sub-organismal, replicator biology. Replicator biology articulates (some of) the mechanisms by which organisms achieve their capacities. We should see organismal and sub-organismal biology as, in some sense, complementary. So the causal capacities of sub-organismal units and the capacities of organisms as natural purposes have ineliminable, mutually dependent explanatory roles to play even in contemporary biology. Developmental biology explains the purposiveness of organisms

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

781

by appeal to the causal capacities of sub-organismal parts and processes, genes, genetic and developmental modules. This is a strictly mechanical form of explanation. But we cannot explain the capacities of sub-organismal components without appeal to the ways in which organisms construed as natural purposes control and regulate them. This is a decidedly unmechanical, teleological form of explanation. We have something very like Kants Antinomy of the teleological power of judgment, recast in explanatory terms. We might call it the Antinomy of teleological explanation: Thesis: All biological phenomena must be explainable in accordance with merely mechanical laws. Antithesis: Some biological phenomena cannot be explained according to merely mechanical laws only (explaining them requires the teleology of natural purposes). This is no cause for rejoicing; an antinomy is a sign of something gone wrong. I believe, though, that we can consistently hold both the thesis and antithesis. But some work needs to be done. The thesis and antithesis look like formal contradictions, so they cannot be consistent, unless there is some equivocation in their respective use of the terms. I think there is an equivocation on the notion of explaining biological phenomena: biological phenomena can be explained severally or collectively. Mechanism, or some contemporary variant, may be able to explain of each organism, how the causal powers of its constituent parts combine to produce the organism in question. But mechanism alone may fall well short of explaining why the kinds of organisms that actually exist do so, or why organismal structure is so regular. Natural purposes, I shall argue, explain biological regularities. I take this challenge up in Section 5. But rst we need to set aside the Kantian conviction that natural purpose is inimical to mechanical law. 4. Natural purpose Kants strategy for resolving the Antinomy of the teleological power of judgment is probably unavailable to the modern day naturalist wishing to resolve the revised antinomy of explanation (Zammito, 2006). Available or not, it is not clear that Kants approach would be particularly palatable to the modern day naturalist, for a couple of reasons. The rst is Kants conviction that purposiveness cannot be demonstrated to be an objective feature of the world. Nevertheless, Kant insists, we must treat teleology and mechanism alike as regulative principles, guiding discursive thought (Guyer, 2005; McLaughlin, 1990). The second is Kants insistence that teleology does not gure in genuine explanation: all explanation is mechanistic. A contemporary naturalist committed to the revised antinomy should rst attempt to demonstrate that the phenomenon of organisms as natural purposes is an objective, natural phenomenon, consistent with the rule of mechanical law. The naturalist should then proceed to show, if possible, that natural purpose has the same explanatory credentials, as does mechanism. This section and the next attempt to demonstrate these things in turn. One of Kants reasons for denying that teleology is an objective feature of the natural world is the inconsistency he perceives between the purposiveness of organisms and the nature of matter. A mechanical explanation, according to Kant, demonstrates that the phenomenon to be explained is wholly a consequence of the nature of matter. But organisms are self-organizing and self-building, and matter, by its nature, is inert.

782

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

However, the possibility of a living matter (the concept of which contains a contradiction, because lifelessness, inertia, constitutes its essential characteristic), cannot even be conceived : : : (Kant, 2000, p. 394) The choice of inertia as the constitutive property of matter, of course, is no accident. Kant is arming Newtons rst law. Indeed, Kant reinterprets Newtons law of inertia as one that holds that all changes in matter have extraneous (external) causes (Watkins, 2001). If Newtons rst law articulates the nature of matter, then matter could not be self-moving, self-organizing or self-building. Kants argument has strong resonances with a conundrum raised by Erwin Schro dinger (1944) in his classic essay What is life?. Living things, according to Schro dinger, embody a paradox. They appear not to be consistent with the most basic principle of physics, and yet they must be. Whereas for Kant the fundamental principles of physics were embodied in Newtons laws, for Schro dinger they are enshrined in thermodynamics. The material world inexorably moves toward a state of maximum entropydisorder. The distinctive feature of organisms, however, is that they spontaneously build order and actively maintain it. In metabolism they convert unstructured matter into highly structured, organic materials. In the process, they concentrate and store energy. Organisms are radically negentropic.6 Self-organizing matter is as problematic for Schro dinger as it is for Kant. Although the paradoxes have similar forms, Schro dingers resolution is starkly dierent from Kants. Schro dinger points out that organisms do not, in fact, violate the second law of thermodynamics. The second law applies to closed systems in which energy is neither added nor lost. The tendency to increase in entropy is a global property of such systems. So long as closed systems as a whole increase in entropy, open sub-systems may consistently decrease in theirs. How do systems decrease entropy locally with systematic regularity? In systems far from thermodynamic equilibrium, where signicant energy gradients occur, it may well be that the most ecient way of dissipating energy is to build ordered structures to nard) cells provide an illustrative examdo the job. The marvellous order of convective (Be ple. Heat applied to the underside of a pan containing a thin layer of oil will set up a dense, nard cells) of uniform size. This stable arrangement of hexagonal convective cells (Be organization is robust; if perturbed it will return to its stable conguration. The explanation of this phenomenon is that this conguration is the most eective means for dissipating energy from the bottom of the layer of uid to the top. These structures are negentropic, but they do not violate the second law. In Schro dingers phrase, they pay their debt to the second law by greatly increasing the entropy around them. Schro dinger concludes, rather triumphantly, that the self-organizing, self-regulating features of living things do not violate the laws of thermodynamics. The spontaneous construction of stable, ordered structures is to be expected in systems far from thermodynamic equilibrium. Rather than violating the second law of thermodynamics either, self-organization is predicted by thermodynamics. Schro dinger contends that the constitutive property of living things is metabolism: organisms maintain their negentropic structures by synthesizing the materials from which they are built (cf. Kant, 2000, p. 371). Maturana, Varela & Uribe (1974) appeal to a similar form of self-organizationautopoiesisin their denition of living things. A living thing is
These features that Schro dinger took to be constitutive of living things are the very properties of organisms as natural purposes that occasioned such disquiet in Kant.
6

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

783

an autopoietic machinean embodied, self-organizing entity that actively maintains its structure. Embodiment here means that the machine synthesises the materials out of which it is built (Boden, 2000). Stuart Kauman endorses this account of the nature of living things, and he explicitly invokes Kants reciprocity of causes in describing the self-organizing, self-sustaining autocatalytic reactions characteristic autopoietic machines: Immanuel Kant : : : saw organisms as wholes. The whole existed by means of the parts: the parts existed because of, and to sustain, the whole. This holism has been stripped of a natural role in biology, replaced with the image of the genome as the central directing agency that commands the molecular dance. Yet an autocatalytic set of molecules is perhaps the simplest image one can have of Kants holism. (Kauman, 1995, p. 69) Kant and Schro dinger both take metabolismself-organization and self-buildingto be the mark of an organism. Kant takes these processes to be the mark of purposiveness. Whereas Kant avers that this purposiveness cannot be an objective principle, Schrodinger argues that it can. But Schro dinger locates self-organization as an objective property, not of matter as such, but of an organized system.7 Self-organization has become one of the most fruitful avenues of research in modern evolutionary biology (Camazine et al., 2003). The mechanisms that underwrite self-organization and the capacities of self-organizing systems are becoming increasingly well understood (Heylighen, 2001). Organisms are complex self-organizing systems that exist on the edge of chaos, neither too sensitive to perturbations to be thrown into chaos, nor too resistant to them to be xed and frozen (Kauman, 1993, 1995). In such systems the whole is the consequence of the causal capacities and activities of its component parts. Yet the capacities and activities of the component parts are the consequences of the adaptive plasticity of the system as a whole. These systems are both cause and eect of their constituent parts. The robust, adaptively plastic, self-organization that marks organisms out as natural purposes is a simple consequence of a natural tendency of matter far from thermodynamic equilibrium. Kant errs in his claim that natural purposes are inconsistent with the nature of matter. We can now locate purposiveness as an objective part of the mechanistic, natural world. : : : teleological behaviour directed toward a characteristic nal state or goal is not something o limits for a natural science and an anthropomorphic misconception of processes which, in themselves, are undirected and accidental. Rather it is a form of behaviour which can be well dened in scientic terms and for which the necessary conditions and possible mechanisms can be indicated. (Von Bertalany, 1969, p. 46) The thesis of the antinomy is true; organisms as natural purposes are susceptible of mechanical explanation. But this naturalization of biological teleology does not particularly promote the project of resolving the revised antinomy. To the extent that these systems-theoretic, approaches to teleology oer a passably mechanistic explanation of teleology, at the same time they weigh heavily against the truth of the antithesis.8 What
I thank Joan Steigerwald and Peter McLaughlin for help here. Ginsborg (2004), p. 64 n. 47, expresses a similar worry, that naturalizing teleology in this way threatens the presumed indispensability of organisms as ends.
8 7

784

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

needs to be shown now is that the antithesis is true, despite the naturalization of teleology: some biological phenomena cannot be explained except by appeal to natural purposes, even though natural purpose itself is mechanistically explicable.

5. Causes, laws and explanations The concept of a law of nature is ambiguous; at least it is compound. It is used variously to cover two distinct but related concepts: governance and systematicity. By governance I mean that the laws of nature govern the occurrences of things. An increasingly prominent view of laws of nature holds that laws derive from fundamental capacities of things, the capacities that constitute their natures. These capacities necessitate their eects. The laws that describe how things with various causal powers are intrinsically disposed to act in virtue of having these powers are the causal laws of nature (Ellis, 2001, p. 206). This view of natural law, it would seem, is quite congenial to Kant. Kants conception of governance reects this idea of laws of nature deriving from the natures of things: : : : the laws of nature depend upon the natures that things have, and it is these natures that govern the activities of substances in determining what happens (Watkins, 2005, p. 406). By systematicity I mean that laws of nature are supposed to be manifested as counterfactually robust empirical regularities. Kant appears to assume the systematicity of laws at least as a regulative principle (Breitenbach, 2006). According to Kant the regularities of the physical world are to be explained by appeal to the governance of laws that constitute the nature of matter. Governance and systematicity of law are intimately associated in the Pure Mechanism of Newton, so admired by Kant. The world is regular precisely because it is governed by a few fundamental, systematic laws. Wherever we see an instance of regularity we are entitled, according to Newton, to infer the action of a particular law, acting on the same initial conditions. Therefore to the same natural eects we must, as far as possible, assign the same causes (Newton, 1963, p. iii). On the Pure Mechanism model of explanation for any eect e, that instantiates regularity R, the factor that explains es occurrence is the same as the factor that explains its instantiation of R. This has been enshrined in the canonical concept of explanation throughout much of the 20th century, the DN model. According to the DN model of explanation, to explain an eect, e, is to subsume it under a general law, R, such that R (in conjunction with some initial conditions) entails the occurrence of e. Nevertheless, governance and systematicity can be dissociated.9 Natural law does not invariably issue in systematicity of eects and the systematicity, or regularity, of eects may not be the consequence of unity of governing natural law. Chaotic and self-organizing systems oer interesting illustrations. Chaos theory gives us an example of governance without systematicity. Chaotic systems are deterministic. Given the laws of nature and precise initial conditions, a given chaotic system will produce its outcome as a matter of nomic necessity. But the laws governing chaotic systems do not gure in anything like systematic empirical regularities. Chaotic systems are characterised by hypersensitivity to initial conditions. A minor variation in initial conditions can result in an enormous dierence in
Nancy Cartwright (1999), for example, argues that the laws governing the workings of the world are not systematic, but make up a patchwork of merely local regularities.
9

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

785

outcome. Consequently, for any given chaotic system, if we know the initial conditions and the laws it is obeying, we get a complete explanation of its evolution. But this gives us no prediction or explanation of what would happen in other counterfactual situations in which the initial conditions are even minutely dierent. Self-organizing systems, on the other hand, oer an example of systematic regularity of phenomena that is not attributable to the unity of governing law. The robustness of self-organizing systems enables them to produce and maintain a trajectory toward a particular endpoint despite signicant dierences in initial conditions and perturbations in the process. It is not the unity of law, or of underlying causes, that determines the systematic regularity of self-organizing systems; it is something else, a structural property of the system as a whole. Where governance and systematicity peel apart in this way, it is natural to suppose that an explanation of the mechanical causes of a particular event is not eo ipso an explanation of some regularity (and vice versa). Kant appears to countenance the important distinction between the mechanical explanation of regularities and the mechanical explanation of singular occurrences. Some phenomena, while explicable mechanistically, are nonetheless contingent with respect to mechanical law (Breitenbach, 2006; Ginsborg, 2001). One source of this contingency may be that some events, while wholly explicable causally, are not likely to be regular. An example might be a particular avalanche or volcanic eruption (chaotic systems in general will have this property). There is nothing about the general nature of matter that explains the specic details of these occurrences. They are highly irregular. What is so puzzling about organisms is that they are both so contingent with respect to mechanical law and also regular. Thus when Kant says that we cannot conceive of how unorganized matter left to its own workings could spontaneously form itself into an organism, his point is that the regularities exhibited by organisms cannot be accounted for in terms of fundamental regularities that characterize the behaviour of matter at the most general level (Ginsborg, 2001, p. 245). Biological regularity is not explicable by appeal to mechanical law. 5.1. Dierence making Intuitively, to explain an event is to single out those conditions that make the dierence between the events occurrence and its non-occurrence: to explain is to identify a dierence maker. Typically the relation between the phenomenon to be explained, P, and its dierence maker, D, is one of invariance (Woodward, 2000). Invariance is a relation between a range of properties {d1, d2, : : :, dn} that D might possess and a range of properties {p1, p2, : : :, pn} that P might possess. The relation di, pi is invariant if and only if it is (i) change relating and (ii) robust. Change relating is meant to capture the idea that a dierence in the value of di brings about a dierence in the value of pi. Robustness is meant to capture the idea that the relation between di and pi is insensitive to a range of other factors extrinsic to di, pi; as Garnkel points out : : : an explanation must have a certain amount of stability under perturbations of its conditions (Garnkel, 1990, p. 448). When the invariance relation holds, we can use it answer questions of the what-if-things-had-been-dierent sort (Woodward, 2000).10 To be in a position to answer such a question of some phenomenon, P, is to be able to explain P.

We need also to be in the position to answer the question: what would have to be the same for things to work out this way again?.

10

786

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

The dierence making approach to explanation reveals how, for some event or process e, that instantiates regularity R, the explanation of es occurrence may dier from the explanation of es instantiating R. An example (misappropriated) from Elliott Sober (1984) illustrates the point. Suppose we observe that all the children in a particular classroom read at the grade three level and want to know why. The appeal for an explanation may be read in either of two ways. We may want to know of all the children severally why each reads at that level. If so we must look to the particular history of each child, and we cite that factor that makes the dierence between her reading at the grade three level, rather than some other. Alternatively, we may want to know why the constitution of the classroom is such that all the children read at that level. Here we advert to those features that make the dierence between this class having this particular composition rather than some other. In Sobers example, the class is constituted this way because there is selection for reading ability; only those students who can read at the grade three level are allowed into the class. This is a case in which the occurrence of an event token e say, Sarahs reading at the grade 3 leveland es instantiating a regularity Rthat all children in the class read at that levelcall for dierent explanations. The occurrence of e and its instantiation of R are sensitive to (stable across) dierent dierence makers.

5.2. Mechanical dierence making and purposive dierence making Using the apparatus of dierence making we can demonstrate that Kants conception of organisms as natural purposes deserves a central, irreducible place in the understanding of evolutionary biology. This requires two steps, each of which exploits a distinctive feature of the invariance account of explanation. In the rst step, we demonstrate the consistency of the thesis and antithesis of our revised antinomy of teleological explanation. In the second, we establish the reciprocal causal relation between an organism and its parts. The rst step, resolving the revised antinomy, avails itself of the fact that for some token event e that instantiates a regularity R, the explanation of es occurrence may be distinct from the explanation of es instantiation of R. Suppose we wish to explain some biological even token e, say the development of some a particular vertebrates eye; e is both an eect of local causes and an instance of a particular (heritable and developmental) regularity, R. There will be a complete microstructural, mechanical explanation which cites all the causal factors in the etiology of e, call these collectively ci, : : :, ck. The occurrence of e is counterfactually dependent upon ci, : : :, ck. If ci had been dierent, for instance, so would the chain of subsequent events leading to e. Yet ci, : : :, ck reliably produces e across a range of extraneous circumstances; ci, : : :, ck is a mechanical dierence maker for e. As a genuine mechanical explanation, this will demonstrate that the occurrence of ci, : : :, ck necessitates the occurrence of e as a matter of law. Kants conception of mechanistic explanation allows that eects of this sort are susceptible, severally, of mechanistic explanations: : : : the claim that organisms are inexplicable on mechanical laws : : : does not itself entail that we could never explain the origin of an individual bird as a lawlike consequence of the precise arrangement of matter to be found in an intact egg (as we might explain the ringing of an alarm clock as a lawlike consequence of its state just after having been wound up). (Ginsborg, 2001, pp. 242243)

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

787

By the same token, the mechanical explanation of e, by appeal to its causal antecedents, ci, : : :, ck, fails to explain why e instantiates R. At this point, Kant cannot appeal to the idea that purpose plays a role in explaining biological regularities in just the way that mechanism explains occurrences severally. For Kant, as we have seen, purposiveness is not an objective feature of the world. Moreover, for Kant, purposiveness does not gure in any genuine natural explanations. But, given the account oered above of purposiveness as a natural phenomenon, and the apparatus of dierence making, we can help ourselves to a solution unavailable to Kant. Organisms exhibit a startling degree of what Garnkel calls redundant causality (Garnkel, 1990, p. 448). There are many alternative causal paths that an organism can exploit in attaining its stable end state. This is just, of course, a consequence of their plasticity. Because of the redundant causality in organisms, R is not counterfactually dependent upon ci, : : :, ck. There are many ways to achieve R and if our particular individual had not taken path ci, : : :, ck to e, plasticity would have ensured some other route to the instantiation of R. So ci, : : :, ck is not a dierence maker for R; plasticity is. So there are two distinct aspects of our target event e to be explained, its occurrence and its instantiation of R, and these are sensitive to dierent dierence makers: the microphysical causes of e and the purposiveness (plasticity) of the organism respectively. Biology, then, has a division of explanatory labour between the mechanical nature of biological processes and their purposiveness. This demonstrates that the thesis of the revised antinomythat all phenomena (severally) have causal explanations that advert to mechanical lawdoes not entail the falsity of the antithesisbiological phenomena, taken collectively, may have non-mechanical, nonnomic, teleological explanations. Thesis: All biological phenomena must be explainable [severally] in accordance with merely mechanical laws. Antithesis: Some biological phenomena cannot be explained [collectively] according to merely mechanical laws only (explaining them requires the teleology of natural purposes). Taken by itself, however, this resolution of the revised antinomy concedes too much to Pure Mechanism, in doing so it falls short of establishing the most important feature of Kants conception of organisms as natural purposes. The consistency of the thesis and antithesis, as I have reconstructed them, does not preserve the idea that an organism is both cause and eect of itself. For all it tells us, organisms may be merely eects of the suite of component processes. There has been no real gain on sub-organismal biology. The second step in rehabilitating Kants conception of the organism involves demonstrating that the suite of causal processes that actually occurs in the development of an organism is also (reciprocally) dependent upon the purposiveness of the organism as a whole. This step also exploits the invariance approach to explanation. As we have seen, the component processes of ontogeny are mechanical dierence makers for the purposiveness of organisms. At the same time, the purposiveness of organisms is the dierence maker for the suite of component causal processes. Uniquely in organisms, mechanism and purpose are reciprocal dierence makers. The phenomena of phenotypic repertoire and phenotypic accommodation demonstrate this. An organism has a broad phenotypic repertoire. Each of its component parts can

788

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

produce a wide range of developmental outputs. The particular output each part produces on an occasion depends on the regulatory relations of the organism as a whole. In phenotypic accommodation the component processes of development are altered by the adaptive, regulatory capacities of the organism as a whole. Should one part of the developing organism be impinged upon by environmental pressure, or mutation, the rest of the developmental system of the organism implements compensatory changes in order to maintain a well functioning, viable organism. Phenotypic accommodation reduces the amount of functional disruption occasioned by developmental novelty (West-Eberhard, 2003, p. 147). Kant appears to anticipate West-Eberhards concept of phenotypic accommodation, to some degree, in his discussion of how the purposiveness of the organism as a whole regulates the component processes of development: If birth defects occur, or deformities come about during growth, certain parts : : : form in an entirely new way, so as to preserve what is there and so produce an anomalous creature (Kant, 2000, p. 372). The important point for our purposes is that the particular suite of causal processes that actually occur within a developing organism only do so because of their regulation by the purposiveness (plasticity) of the whole organism. We would have no explanation of why the mechanisms of development play the causal roles they actually play in the production and maintenance of a viable organism typical of its kind, without the purposiveness (plasticity) of organisms. This may be part of what Kant is adverting to by his claim that the organism is contingent with respect to mechanistic causes (Ginsborg, 2001); that is, : : : that nature, considered as a mere mechanism, could have formed itself in a thousands different ways without hitting upon the unity in accordance with such a rule [the structure of a particular organism] (Kant, 2000, p. 360). There is, then, an invariance relation between the plasticity of the organism as a whole and the causal powers of its component processes. It is (i) change involving: a change in the overall structure of the organism would occasion a change in causal properties of the components; and (ii) robust: a developmental component has its particular causal powers despite internal and external perturbations. This invariance relation is the exact reciprocal of the invariance relation between the causal powers of an organisms component processes and its plasticity. So, mechanical and purposive explanations in biology are mutually dependent. Each kind of dierence maker (mechanical and purposive) has the other kind of dierence maker as a dierence maker. This is the sense in which organisms are causes and eects of themselves. 6. Conclusion It goes without saying that the organismal perspective has fallen largely out of favour in post-Darwinian biology. The very idea of explanation by appeal to the purposes of organisms has come to be seen as problematic, a metaphysical delusion. Besides, orthodox opinion in contemporary evolutionary biology holds that there is no need to extend to organismsmuch less their purposivenessan explanatory role if sub-organismal causal mechanisms are explanatorily adequate. To the extent that Kants conception of organisms as natural purposes has received any consideration from contemporary biology at all, it has been largely dismissed as an irrelevance. I have attempted to do two things here. The rst is to emphasize that the properties of organisms which, according to Kant, mark them out as natural purposes, are observable. It does not require immersion in the critical philosophy to see that organisms

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

789

are self-organizing, self-building adaptive systems. If recent evolutionary developmental biology is to be believed, the purposiveness of organisms is a contributing cause of both ontogeny and adaptive evolution. The second is to carve out an explanatory role for organismal purposes that is consistent with the modern commitment to mechanism. Here, contemporary biologyand contemporary metaphysicsencounter precisely the problem that motivated Kants Antinomy of the teleological power of judgment: teleology appears to be inconsistent with mechanism. Mechanical causes explain; purposes cannot. Moreover, explanation by purpose appears to be otiose. If each natural phenomenon can be fully explained by adverting to its mechanical causes, there is no unexplained residue for which purposes are needed. As I see it, anti-teleological scruples such as these are motivated by the lack of a plausible model for how purposes could explain. I have tried to sketch out such a model here using the notion of invariance explanation, which yields two results, one fairly specic, the other quite general. The specic result is that there is an explanatory invariance relation between the purposiveness of an organism and the causal capacities of its parts that directly mirrors the invariance relation between the causal capacities of an organisms parts and its overall purposiveness. Purpose and mechanism are reciprocal causes. Given that, even if the occurrence of each biological event has a complete mechanical, sub-organismal explanation, it does not follow that there is no explanatory role for the purposiveness of organisms to play. Purposiveness explains the regularity of these occurrences, and this is crucial for the understanding what makes adaptive evolution adaptive. The more general result is that, on the invariance model, purpose explains just like any other kind of cause. Acknowledgements I wish to thank Peter McLaughlin for comments on an early draft, Marcel Quarfood for engaging discussion on these issues, and audiences at IHPST, Toronto and the Consortium for the History and Philosophy of Biology workshop in Montreal. I would particularly like to thank Joan Steigerwald, rst for the invitation to contribute and then for patience and acumen as editor. References
Amundson, R. (2001). Development and selection: On the lack of a common ground. In S. Orzack, & E. Sober (Eds.), Adaptation and optimality (pp. 303334). Cambridge: Cambridge University Press. Ayala, F. (1970). Teleological explanations in biology. Philosophy of Science, 37, 115. Bell, G. (2000). Selection: The mechanism of evolution. New York: Chapman Hall. Bergman, A., & Siegal, M. (2002). Evolutionary capacitance as a general feature of complex gene networks. Nature, 424, 549552. Boden, M. (2000). Autopoiesis and life. Cognitive Science Quarterly, 1, 117145. Boyd, R., & Richardson, P. (2004). Not by genes alone. Cambridge, MA: MIT Press. Breitenbach, A. (2006). Mechanical explanation of nature and its limits in Kants Critique of judgment. Studies in History and Philosophy of Science, this issue. doi:10.1016/j.shpsc.2006.09.001. Broad, C. D. (1925). Mind and its place in nature. London: Routledge & Kegan Paul. Camazine, S., Deneubourg, J.-L., Franks, N. R., Sneyd, J., Theraulaz, G., & Bonabeau, E. (2003). Selforganization in biological systems. Princeton: Princeton University Press. Cartwright, N. (1999). The dappled world: A study in the boundaries of science. Cambridge: Cambridge University Press.

790

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

Cornell, J. J. F. (1986). Newton of the grassblade? Darwin and the problem of organic teleology. Isis, 77, 405421. Cummins, R. (2002). Neo-teleology. In A. Ariew, R. Cummins, & J. Perelman (Eds.), Functions: New essays in the philosophy of biology and psychology (pp. 157173). Oxford: Oxford University Press. Darwin, C. (1972). The origin of species (6th ed.). Toronto: Dent. (First published 1859) Dawkins, R. (1982). The selsh gene. Oxford: Oxford University Press. Dawkins, R. (1986). The extended phenotype. Oxford: W. H. Freeman. Ellis, B. (2001). Scientic essentialism. Cambridge: Cambridge University Press. Fisher, R. A. (1930). The genetical theory of natural selection. New York: Dover. Fox Keller, E. (2000). The century of the gene. Cambridge, MA: Harvard University Press. Garnkel, A. (1990). Reduction. In R. Boyd, P. Gaspar, & J. D. Trout (Eds.), The philosophy of science (pp. 443 459). Cambridge MA: MIT Press. (First published 1981) Ghiselin, M. (1993). Darwins language may have been teleological, but his thinking was a dierent matter. Biology and Philosophy, 9, 482492. Gibson, G. (2002). Getting robust about robustness. Current Biology, 12, R347R349. Ginsborg, H. (2001). Kant on understanding organisms as natural purposes. In E. Watkins (Ed.), Kant and the sciences (pp. 231258). Oxford: Oxford University Press. Ginsborg, H. (2004). Two kinds of mechanical inexplicability in Kant and Aristotle. Journal of the History of Philosophy, 42, 3365. Gould, S. J., & Lewontin, R. C. (1979). The spandrels of San Marco: A critique of the Panglossian paradigm. Proceedings of the Royal Society, Series B, 205, 581598. Greenspan, R. J. (2002). The exible genome. Nature Reviews Genetics, 2, 383387. Griths, P. E., & Gray, R. (2001). Darwinism and developmental systems. In S. Oyama, P. E. Griths, & R. Gray (Eds.), Cycles of contingency (pp. 195218). Cambridge, MA: MIT Press. Guyer, P. (2005). Organisms and the unity of science. In P. Guyer (Ed.), Kants system of nature and freedom: Selected essays (pp. 86111). Oxford: Oxford University Press. Haynes, R. (1988). Biological context of DNA repair. In E. C. Friedberg, & P. C. Hanawalt (Eds.), Mechanisms and consequences of DNA damage processing (pp. 577584). New York: Alan Liss. Herschel, J. (1987). A preliminary discourse on the study of natural philosophy. Chicago: University of Chicago Press. (First Published 1830) Heylighen, F. (2001). The science of self-organization and adaptivity. In L. D. Kiel (Ed.), Knowledge management, organizational intelligence and learning, and complexity. The Encyclopedia of Life Support Systems (EOLSS). www.eolss.net. Hull, D. (2003). Darwins science and Victorian philosophy of science. In J. Hodge, & G. R. Radick (Eds.), The Cambridge companion to Darwin (pp. 168191). Cambridge: Cambridge University Press. Jablonka, E., & Lamb, M. (2005). Evolution in four dimensions. Chicago: University of Chicago Press. Kant, I. (2000). Critique of the power of judgment (P. Guyer, & E. Matthew, trans.). Cambridge: Cambridge University Press. (First published 1793) Kauman, S. (1993). The origins of order: Self-organization and selection in evolution. Oxford: Oxford University Press. Kauman, S. (1995). At home in the universe: The search for the laws of self-organization and complexity. Oxford: Oxford University Press. Lewontin, R. C. (1978). Adaptation. Scientic American, 239, 212230. Mameli, M. (2004). Non-genetic selection and non-genetic inheritance. British Journal for the Philosophy of Science, 55, 3571. Maturana, H., Varela, F., & Uribe, R. (1974). Autopoiesis: The organization of living systems, its characterization and a model. Biosystems, 5, 187196. Maynard Smith, J. (2000). The concept of information in biology. Philosophy of Science, 67, 177194. Mayr, E. (1988). The multiple meanings of teleological. In idem, Towards a new philosophy of biology: Observations of an evolutionist (pp. 3866). Cambridge: Harvard University Press. McLaughlin, P. (1990). Kants critique of teleology in biological explanation. Lewiston, NY: Mellon. McLaughlin, P. (2000). What function explains. Cambridge: Cambridge University Press. Moss, L. (2002). What genes cant do. Cambridge, MA: MIT Press. Newman, S., & Muller, G. (2001). Epigenetic mechanisms of character origination. In G. Wagner (Ed.), The character concept in evolutionary biology (pp. 549579). New York: Academic Press.

D.M. Walsh / Stud. Hist. Phil. Biol. & Biomed. Sci. 37 (2006) 771791

791

Newton, I. (1963). Sir Isaac Newtons mathematical principles of natural philosophy, Vol. 2: The system of the world (A. Motte, trans.; revised by F. Cajori). Berkeley: University of California. (First published 1729) Oyama, S. (1985). The ontogeny of information: Developmental systems and evolution. Durham, NC: Duke University Press. Oyama, S., Griths, P., & Gray, R. D. (Eds.). (2001). Cycles of contingency: Developmental systems and evolution. Cambridge, MA: MIT Press. Sarkar, S., & Gilbert, S. (2000). Embracing complexity: Organicism for the 21st century. Developmental Dynamics, 219, 19. Schmalhausen, I. I. (1949). Factors of evolution. Philadelphia: Blakeston. Schrodinger, E. (1944). What is life? New York: Dover. Sober, E. (1984). The nature of selection. Chicago: University of Chicago Press. Von Bertalany, L. (1969). General systems theory. New York: George Barziller. von Dassow, G. E., & Munro, E. M. (1999). Modularity in animal development and evolution: Elements of a conceptual framework for evodevo. Journal of Experimental Zoology (Mol. Dev. Evol.), 285, 307325. Von Dassow, G. E., Meir, M., Munro, E. M., & Odell, G. M. (2000). The segment polarity network is a robust developmental module. Nature, 406, 188192. Waddington, C. H. (1957). The strategy of the genes. London: Allen & Unwin. Wagner, A. (2005). Robustness and evolvability in living systems. Princeton: Princeton University Press. Watkins, E. (2001). Kants justication of the laws of mechanics. In E. Watkins (Ed.), Kant and the sciences (pp. 136159). Oxford: Oxford University Press. Watkins, E. (2005). Kant and the metaphysics of causality. Cambridge: Cambridge University Press. West-Eberhard, M. J. (2003). Developmental plasticity and evolution. Oxford: Oxford University Press. West-Eberhard, M. J. (2005a). Phenotypic accommodation: Adaptive innovation due to developmental plasticity. Journal of Experimental Zoology, 304B, 610618. West-Eberhard, M. J. (2005b). Developmental plasticity and the origin of species dierences. Proceedings of the National Academy of Sciences in the United States, 102, 65436549. Woodward, J. (2000). Explanation and invariance in the special sciences. British Journal for the Philosophy of Science, 51, 107254. Zammito, J. (1991). The genesis of Kants Critique of Judgment. Chicago: University of Chicago Press. Zammito, J. (2006). Teleology then and now: The question of Kants relevance for contemporary controversies over function in biology. Studies in History and Philosophy of Science, this issue. doi:10.1016/j.shpsc.2006. 09.008.