EVOLUTIONARY THEORY OF KNOWLEDGE: PHYLOGENETIC RELATION BETWEEN REPRESENTATION AND OBJECT

ADRIANNA WOZNIAK

“Plato says in "Phaedo" that "our necessary ideas" arise from the preexistence of the soul, are not derivable from experience — read monkeys for pre-existence” Ch. Darwin, 1837.

Introduction Enigma of the origin of synthetic a priori knowledge For Aristotle, perception was integral to his general theory of causality, which was conceived as a process where one and the same form, originally present in the external object literally comes into the mind (R. Adams, 1975:73). This implies, on the one hand, intuitive transcendental realism: if a form is present in mind, the object from which this form comes from must exist too (or at least have existed); consequently, if a perceptual representation1 exists, its object - its source - must exist too. On the other hand, this transmission-like conception of perception and of causality implies an ante rem position asserting that nihil est in intellectu, quod non fuerit in sensu: if a representation of a snake is present in mind, we cannot know whether the real snake it refers to is venomous or not before we experience it, and if we know that the snake is venomous this is because we have experienced it. Nevertheless, perception does not explain the origin of all knowledge. The enigma starts with the observation that though all knowledge is triggered by experience, all knowledge does not derive from it: “But though all our knowledge begins with experience, it does not follow that it all arises out of experience” (I. Kant, 1980:B1). For example, sensory experience of an event triggers, but does not originate the idea that “every change must have its cause”; the idea of causality implies a necessity that cannot be derived from experience of two events2; what is more, it is present in mind before the experience takes place and determines it. Those necessary ideas, or representations of objects that are neither entirely engendered nor explained by the individual, sensory, experience of the object, Kant called synthetic a priori knowledge (SAK). Since SAK is not derivable from the causal perceptual relation, it is not entirely acquired through sensory experience of the external objects, and thus we cannot prove whether SAK is related to them, or whether the objects present in our mind 1

We use the term representation as designating the way by which an object is given to the subject (Kantian Vorstellung). 2 What was already showed by D. Hume, 1739:T 1.3.14).

1

exist outside it and are something more than the objects of our subjectivity. Thus, contrary to Aristotle, Kant argues: “however clearly we may be conscious of our representation of these things, it is still far from certain that, if the representation exists, there exists also the object corresponding to it” (I. Kant, 1980:A371-372). This is what led Kant to the position of transcendental idealism. If a priori synthetic knowledge is not acquired (since the only mode of acquisition is ontogenetic experience), it is immanent to the soul. Yet, how did SAK come into the mind? Kant mentions Plato’s myth, which in the voice of Socrates explains that before being reincarnated in body, the soul (that contrary to the body does not disintegrate) migrates to Pleroma, the realm of the pure Forms, and contemplates it. Consequently, if we are bearers of a priori ideas (e.g. idea of equality), this is because they exist in Pleroma, and because our souls have been acquainted with them prior to our births (Plato, 1989:105d). This would explain why even though in occurrences of our bodily life experience we never perceive any perfectly equal things, we can estimate the relation of equality between, for example, two pieces of wood; since knowledge cannot be obtained through any individual, ontogenetic experience, it must be a kind of recollection of what the soul experienced in Pleroma. Of course, reincarnation was rejected by Kant. According to the latter, if SAK is not acquired through the empirical experience, it must be given by God, source of SAK’s origin and validity; from this perspective, tercium non datur est.

Evolutionary Theory of Knowledge Nonetheless, a third solution comes with the theory of evolution, destroying the concept of a priori: the knowledge of object that precedes its individual experience is a priori in ontogenetic perspective but takes a posteriori character in the phylogenetic one, being acquired through the evolutionary process of natural selection (K. Lorenz, 1981:101). The latter is considered as a “mechanism thanks to which external cause is transformed into effect” (R. Lewontin, 2003:118-120), an asymmetrical process where « the environment brings about an organic change exactly in its own image » (P. Godfrey-Smith, 1996:86). Thus, natural selection designates a phylogenetic relation between cause (factor from selective environment) and effect (organism). This implies intuitive transcendental realism situated in a phylogenetic perspective: if an organism is bearing a given form, the selective factor responsible for its origin must exist too (or at least have existed). Consequently, if an organic form exists, its environmental cause must exist too. For Aristotle, perception was a mode in which the external objects can causally affect representations; for Evolutionary Theory of Knowledge (ETK), natural selection instantiates causality where one and the same form, originally present in the external object is selected by eliminatory action of the very object among variations available in the genetic pool. In this two step process of evolution by natural selection (variation, elimination), the external object is transmitted into the organism. The esse est percipi assertion is not convincing given that the ability itself to perceive the world results from the selective action of the latter on the knowing subject (R. Meyers, 1990:112). If monkeys are able to leap from one branch to another, their space cognition must correspond with space properties of the world (Fr. Jacob, 1981). Adaptation explains the correspondence between, in Heraclitus’

2

expression, internal and external logos, the origin and the validity of SAK, survival and reproduction being its criterion3. Before ETK, the fact that there is nothing in representation, which does not come from sensory, individual experience, except the senses- nihil est in intellectu, quod non fuerit in sensu, excipe: nisi ipse intellectus (G.W. Leibniz, 1975)- was problematic. Within ETK the cognitive apparatus constitutes an adaptation, evolutionarily gained knowledge: “the way in which adaptations evolve should then be seen as a way of gaining knowledge” (H. Plotkin, 1995:229). Adaptation resulting from the evolution by natural selection is a process through which information about the environment is literarily incorporated to (W.W. Bartley, 1987) or assimilated by an organism (K. Lorenz, 1975:12; cf. G. Vollmer; H. Kornblith). Thus, an ante rem position is defended, asserting that there is nothing in the a priori representation, which does not come from phylogenetic causal experience. If a particular kind of representation of a snake is available, we do know that the real snake it refers to can be venomous before we experience it in the evolutionary past of our species. For Kant, a priori (innate, necessary, universal, non modifiable) was opposed to learned (a posteriori), given that learning was the only known mode of acquisition of knowledge. With the theory of evolution, the innate/acquired dichotomy has been transformed (A. Ariew, et al. forthcoming). The process of adaptation is another way knowledge can be entrenched in the knowing, evolving system, natural or artificial. Innate do not contrast any more with learned, but rather with acquired (E. Sober, 1998) and is perfectly modifiable during the phylogeny. In evolutionary context, innate still implies universal, but in the sense of species-specific feature, which is present in each normally developing individual of a given taxon4. “Plato says in " Phaedo" that "our necessary ideas" arise from the preexistence of the soul, are not derivable from experience — read monkeys for pre-existence” (Ch. Darwin, 18375).

Phylogenetically Acquired Representation (PAR) Within the Evolutionary Theory of Knowledge, the notion of synthetic a priori knowledge (knowledge of the object, which was not entirely acquired by learning, by ontogenetic experience of this object) will correspond to the notion of Phylogenetically Acquired Representation (PAR), founded on the parallel between the notion of representation and of adaptation: Representation

Adaptation

An (a set of) internal state(s) of the A (a set of) hereditary (partly carried agent by open genetic p r o g r a m6) property(ies) of the agent

3

According to the better-safe-than-sorry argument (Stephens, 2001), it is possible that natural selection favours exaggerated reflexes. For example, it is better to have an innate tendency to avoid all snakes, even those that are not venomous, than to risk contact (trial/error learning) with a snake that may, or may not, be venomous. Cf. Sober, 1994. 4 For Locke some ideas were not innate precisely because they were not present in young and wild individuals (Locke, 1690). 5 This is what Charles Darwin wrote in M Notebooks in 1837, at the age of 28. Quoted in Vollmer, Gerhard. 2005. How is it that we can know this world ? New arguments in Evolutionary Epistemology. Darwinism and Philosophy. Hosle, Vittorio ed. University of Notre Dame Press, p. 259. 6 A closed genetic program would be a sufficient one to develop the phenotype, without the recourse of environmental factors. This concept is of course abandoned. The notion of open genetic program includes the

3

that holds a relation of reference

that results from a causal phylogenetic relation

toward certain objects in the external toward an external factor. world. The representation of the object, as present in the mind, does not entirely derive from sensory, individual experience of this object.

The contribution of the sensory, individual experience of this factor is not sufficient for the trait-adaptation to develop.

PAR as adaptation. PARs are (a set of) features of the organism carried by an open genetic program that result from a causal phylogenetic relation to factors from the selective environment. The form of PARs is, thus, not entirely determined by individual experience of the environmental factor. Though all our knowledge becomes available only if triggered by an external factor7, not all knowledge arises out of experience (Paraphrase of I. Kant, 1980:B1). Evolving organisms benefit from the combination of phylogenetic and ontogenetic learning8; knowledge of object as present in mind is built not only from learned components, but depends also on innate, “hard-wired” ones. PAR as representation. Phylogenetically acquired features have representational status, because adaptations (e.g. escape behaviour) corresponding to an environmental factor (e.g. snake), do not derive and cannot be fully explained, by the ontogenetically acquired experience of this factor. The ontogenetical exposure to snakes is not sufficient to acquire the escape behaviour that is triggered once the individual senses a snake. The reason for which individuals of species S fly snakes is not an ontogenetically acquired belief of these individuals, but precisely a PAR, the meaning of a snake being acquired through the phylogenetic experience of S. Natural selection is a process of discriminating sampling occurring when the individuals do not reproduce because their traits do not fit their environment. The probability of individuals to contribute to the next generation depends on their fitness. In Modern Synthesis, natural selection designates a cause/effect relation, whereby the environment (as a fitness value fixed by experimenter) instantiates the cause and the organism instantiates the effect. This causal and externalist characteristic of natural selection guarantees that the main criterion of representation is fulfilled, namely the presence of the causal relation from object to representation. Thus, PAR is every feature that constitutes an adaptation, i.e. resulting from the discriminating process of natural selection. Since the latter can act only on what is heritable, and what is heritable is genetic, a structure that constitutes an adaptation must be fact that the phenotypic development requires a contribution of genes and non-genetic environment (Mayr, 1974:651). 7 “ The blueprint contained in the genome requires innumerable environmental factors in order to be realised in the phenogeny of structures and functions. During his individual growth, the male stickleback may need water of sufficient oxygen content, copepods for food, light, detailed pictures on his retina and millions of other conditions in order to enable him, as an adult, to respond selectively to the red belly of rival. Whatever wonders phenogeny can perform, however, it cannot extract from these factors information which simply is not contained in them, namely, the information that a rival is red underneath”. (K. Lorenz, 1966:37). Cf. A. Ariew ; cf. Platon, 1991, Platon’s (Socrates’) methods of revealing by questioning (a slave boy); cf. also G.W. Leibniz, 1975:49-50 “If there were veins in the block which indicated the figure of Hercules rather than other figures, this block would be more determined thereto, and Hercules would be (…) in a fashion innate in it, although it would be necessary to labour to discover these veins, to clear them by polishing and by cutting away what prevents them from appearing. Thus it is that ideas and truths are innate in us, as inclinations, dispositions, conditions, or natural potentialities”. 8 “Without sensibility no object would be given to us, without understanding no object would be thought. (…) Only through their union can knowledge arise”. (I. Kant, 1980 :B/75-B/76, A/51-A/52. p. 812-813)

4

(partially) innate9. There are three conditions for a feature F to be considered as representing x: • F must enter the state S if x occurs, e.g. trigger escape behaviour in the presence of a sensory experience invoking a predator; F must be an adaptation: • the property of F to enter the state S if x must be the cause thanks to which F was retained in the discriminating process of natural selection • F must be underpinned by the open genetic program (innate to some extent) The evidence and the measure for PARs is the operation of removing empirically derived elements: “For if we eliminate from our experience everything which belongs to the senses, there still remain certain original concepts and certain judgments derived from them, which must have arisen completely a priori, independently of experience, inasmuch as they enable us to say, or at least lead us to believe that we can say, in regard to the objects which appear to the senses, more than mere experience would teach – giving to assertions true universality and strict necessity, such as mere empirical knowledge cannot supply” (I. Kant, 1980:761-A2). It gave the basis for input/output model enabling empirical investigation of innateness: if we take sensory experience as the input and behavioural response of the individual (e.g. escape behaviour) as the output, it turns out that the latter contains more information about the external stimuli that triggered it (e.g. a snake), than provided or acquired by the individual sensory experience of it. From the output we substract the contribution of the experience of the external triggering stimulus; we thus obtain the contribution brought by PARs10. PARs acquire meaning via the phylogenetic process of natural selection. The latter, instantiating a causal relation (E. Sober, 1984), guarantees the relation of correspondence and confers on PARs the status of knowledge. Phylogenetically acquired meaning is a natural and objective property, determined by natural selection. PARs are instantiated by innate features, which constitute adaptations, enable the ontogenetic experience of an object and generate conceptual representations. For example, “the fleshy water-conserving cactus stem constitutes a form of knowledge of the scarcity of water in the world of the cactus” (H. Plotkin, 1995). The wing pattern of some butterflies looks like the eyes of predators of predators of those butterflies, which often tricks the latter. The reason for it is not an idea that those butterflies would have, but natural selection (D. Dennett, 1983). Once this variant available in the genetic pool, it was retained in the process of discriminate sampling. This is why ocella are not observed in populations of butterflies predated by almost blind bats. Meaning can be acquired via two processes: that of ontogenetic learning, at work in individual development, and that of natural selection, at work in the phylogeny of species. Ontogenetically Acquired Representation (OAR) constitutes learnt belief (never innate), deriving its conceptual content from learning, and acquires meaning during ontogeny when the individual learns what an object indicates; OAR is the cause of behaviour, a voluntary representation (Fr. Dretske, 1999). For example, a bird learns that Monarch is indigestible leading to avoidance behaviour. Reflexes over which the individual has no control are not OARs- this is the question of agency: I have reason to do something but this is not the reason I’m doing it; OARs must be the reason and the cause of actions (Fr. Dretske, 1999). Since the acquisition time for OARs is ontogeny and they are not hereditary, OARs are beyond the scope of selective explanations (natural selection acts only on what is hereditary). 9

Not every innate trait has to be an adaptation. Cf. W. V. O. Quine, 1964 ; N. Chomsky, 1975.

10

5

However “human and non human animals do some things not because they have beliefs and desires but because they have innate representations of phylogenetic origin. Behaviours about which I am thinking are intentional behaviours because they depend on representations. But representations do not depend on propositional attitudes having a conceptual content; they are sensory representations having a non-conceptual content. I will qualify those intentional behaviours as involuntary: an intentional involuntary behaviour depends on a representation having a non-conceptual content” (P. Jacob, 1997:285-286). Selective explanations are applicable for PARs referring to synthetic a priori knowledge, which is not, either the cause of behaviour, or learned via ontogenetic experience: “But our understanding is not, by its representations, the cause of the object (…) anymore than the object is the cause of understanding’s representations (…). Thus the pure concepts of understanding must be neither abstracted from sense impressions, not should they express the receptivity of representations through senses, but in truth, have their origin in the nature of the soul [innate A.W.] without however being caused by the object or producing themselves the object”11. Synthetic a priori knowledge does not refer to rational process, available to conscious control (K. Lorenz, 1981). “Some intelligent actions, after being performed across several generations, become converted into instincts and are inherited, as when birds on oceanic islands learn to avoid man. These actions may then be said to be degraded in character, for they are no longer performed through reason or from experience”. (Ch. Darwin, 1871) PARs are all involuntary behaviours that are not conscious intentions. Here are few examples: reflexes, motor responses to a sensory stimulus without the help of consciousness, like tropisms of some plants and of less complex animals (e.g. the aptitude to follow light or temperature gradient of tick); smile or frown; fear of snakes; vertigo, present in newborn baby (I. Eibl-Ebesfeld, 1989); causal, inductive12, spatiotemporal (K. Hahlweg and C.A. Hooker, 1989) cognition; Theory of Mind, aptitude to consider others as having a mind (S. BaronCohen, 1995), even when others belong to a prey or predator species (S. Mithen, 1996); natural kinds cognition (E. Spelke, 1994); folk mathematics (S. Dehaene, 1997); folk biology (cross cultural studies of categorisation (Medin, D and S. Atran, 1999) showed the universality of folkbiological classifications or “default” taxonomy” within the Human species, are particular ways that people conceptualise living kinds, according to the features relevant to theirs usual interactions (e.g. beneficial/noxious, edible/inedible, predator/prey, etc.); essentialism (S. Gelman, and L. Hirschweld, 1999); folk physics (R. Baillargeon, 1986), e.g. capacity to perceive an object’s movement as continue in spite of subject’s jerky movements data, would be common to all land vertebra; specialised neural networks, called modules (J. Fodor, 1983; J. Tooby and C. Leda, 1989); etc.

What PARs are not. Theory of intentionality and adaptationist explanations do not refer to the same register Consciousness is not a criterion for representational and intentional properties. These include, intentional involuntary properties in addition to intentional voluntary properties (P. Jacob, 1997). Evolutionarily inspired Artificial Intelligence showed that evolving virtual agents can act and stay in a viable state without explicit representations, neither of the world 11

letter for Marcus Herz, 21th of February 1772 Kant, Emmanuel. 1980. Œuvres philosophiques, I Des premiers écrits à la Critique de la raison pure. Edition Gallimard. p. 691-692. 12 “Our inductive inferences are tailored to the causal structure of the world” (H. Kornblith, 1993:91); “Creatures inveterately wrong in their inductions have a pathetic but praiseworthy tendency to die before reproducing their kind” (W.V. Quine, 1969:126).

6

nor of themselves in that world (P. Wallis, 2004). Robots act thanks to the causal relation between sensors and actuators (R. Brooks, 1991) without having consciousness, memory or any other internal state encoding an external state. In the beginning of phenomenology, the notion of intentionality, introduced by Brentano (1995) referred to the property of consciousness to be about something. Since then, behaviour joined the rank of representational and intentional properties, action in the world (knowledge of how) being evolutionarily anterior to explicit knowledge of the world, conscientious or expressed semantically (knowledge that). For example, Pushmi-Pullyu representations (P-Ps, Millikan, 2004) are not necessarily propositional. Pushmi, the descriptive side and pullyu the directive side, at the same time represent a fact and drive an appropriate action. Representations are not necessarily mental. The theory of intentionality (the theory of teleosemantic) creates the notion of direct proper function (based on the teleological notion of function borrowed from biology) through which the notions of adaptation and of function are blended, where the normal function of a factor x is to indicate another factor, y, the property for which x was selected. However, the theory of intentionality cannot constitute an aspect of adaptationism (contrary to D. Dennett, 1983:351). The reason for this is that properties, to be considered as intentional, must be ontogenetically acquired (F. Dretske, 1999); properties to be explained in selective terms must be hereditary. Thus, intentionality is a versant of theory of reference, but does not refer, either to the same kind of representational properties (OARs), or to the same kind of explanation that Evolutionary Theory of Knowledge does (PARs).

What PARs are not. Non-commutative triad: a priori, innate and analytic Until Leibniz, a priori is commutative with analytic and innate, given that the proof of innate knowledge “does not depend upon examples, nor consequently upon the testimony of the senses, although without the senses it would never have occurred to us to think of them” (G.W.Leibniz, 1975:45), but only upon internal, analytic considerations. For example, the belief that [p, where p means all bachelors are unmarried] is considered as innate and is justified a priori, by its analycity, without recourse to the experience of every bachelor. Obviously, an empiricist’s counterargument for it would be that innate cannot be analytic, since the latter consists in “a bare explication or understanding of the terms” (J. Locke, 1975:58), and that it is evident that we learn the terms and their meaning and are not born with them. In response to this argument Leibniz explained (G.W. Leibniz, 1975:58-59) that the core of a priori constitute not the names, arbitrary and conventional, but “original characters stamped on the mind”, of which names are a simple expression. Leibnizian a priori/innate- that does not spring solely externally, from the sensory, learnt experience (G.W. Leibniz, 1975:53-54)- becomes synthetic a priori in Kant. With the latter, a priori and analytic was clearly distinguished from synthetic a priori, true in the virtue of an external cause: extensive judgments which “add to the concept of the subject a predicate which has not been in any wise thought in it, and which no analysis could possibly extract from it » (I. Kant, 1980 : A7/B11:767). The question of analicity is another question. If a priori means non a posteriori, A knows a priori that p only if the belief of A that p is not justified by ontogenetic experience and is in need of another explanation. Analycity, proving at best a coherence, is unsatisfactory for the objectivist, referring to an external causes criterion (W.D. Hart, 1975:108). However, innate, contrary to analytic, implies a genesis (St. Stich, 1975:17), a causal origin, and, within the Evolutionary Theory of Knowledge, an adaptive genesis. Within adaptationism, evolutionary cause is at the same time formal (responsible for the origin) and final (relating to the purpose) (D. Dennett, 1983). The 7

evolutionary raison d’être of (partial) innateness would be the vital advantage it provides to ensure appropriate and rapid behavioural responses. If a representation were to relate to a snake only if the latter were present at the moment of content’s acquisition (as Dretske, 1999 wants), the meaning would be available after a contact with the snake, and maybe too late. To be efficient, a reaction must be triggered as soon as the organism perceives the snake, however dimly, which shows danger recognition either prior from experience, or from a limited experience13. Innate reactions mean, on the whole, quick and appropriate reactions. It is clear, that if an organism faces its (starving) predator, the more quickly escape behaviour is triggered, the better. Innateness explains how it is possible “that human beings whose contacts with the world are ephemeral, particularized, limited, are able however to have so much knowledge” (B. Russell, 1948:5) and “how we can acquire such extensive knowledge systems given the fragmentary and restricted character of our experience” (N. Chomsky, 1991:13). The adaptationist argument explains why we are in possession of innate modes of knowledge “which perhaps serve only to give coherence to our sense-representations” (I. Kant, 1980:A2), it does not answer however the question whether innateness of a belief, and its adaptive origin affects in some way the chances that it is true and is not a misbelief. Indeed, here by innateness “we have still not explained what justifies a priori knowledge in even sketchy way we said that experience justifies a posteriori knowledge. We have not yet informatively specified a genuine variety of justification peculiar to a priori knowledge, as experience is peculiar to a posteriori knowledge (…). After all, “why could not A have been born believing that 2+2=69?” (W.D.Hart, 1975:109). To enable organisms to coordinate their actions with environmental conditions (since the function of cognition is to enable agents to deal with environmental demands, P. Godfrey-Smith, 2002), representational properties are constrained by the requirements of viability; they must be in a position to cope with the external world, and for this they must be truth at least to a certain degree14. However, one can argue that survival guarantees solely the viability and phylogenetic success of organisms bearing an adaptation/belief, though not the truth of the latter. The question of the possible evolvability of misbelief will remain open (D. Dennett and R. McKay, forthcoming), since the more direct goal of our dissertation is to examine the possibility of Modern Synthesis to seize adaptationist, causal, phylogenetic relations in order to answer the question of how the external objects can causally affect representations.

13

Cf. E. Sober’s argument for evolution of innate beliefs (originally conceived to explain, by evolutionary advantage of non solipsist beliefs, why in human mind the belief in the existence of the external world evolved): “It isn’t that the solipsist can never figure out that tiger images and dead antelope images have something interesting in common. Perhaps repeated experiences of each will cause a solipsistic organism to realize that behaviours appropriate to the former also are appropriate to the latter. But this discovery requires a long run of samples, one that predatory tigers may summarily nip in the bud. The great thing about the non solipsist is that its belief in tigers provides it with a category that even first exposure to a dead antelope may trigger. The first dead antelope is probably an interestingly novel type of experience for the solipsist, but is more likely to a sign of danger to the non solipsist. This difference can make all the difference” (E. Sober, 1994:36). 14 For instance, diurnal organisms perceive first black on white and recognize the features as curved according to their brightest side. That way to recognizes convexity is due to the fact that the sun light always came from above (I. Eibl-Ebesfeld, 1989). Another thought experiment (S. Dehaene, 1987) suggests that some rudimentary mathematics are PARs (adaptations) common to certain organisms because it was their phylogenetic be or not to be and this implies their truth: if to quench their thirst individuals of species A needed to reach limited sources of water (a cave with the only spring in the area), they could survive only if they knew that if two individuals from predating species B enter the cave, and if afterwards only one of them goes out, there were still one predator in, ready to savour, in addition to nice cold water, warm meat. Not to be eaten don’t make mistakes in the simple subtraction: 2-1=1. Lorenz was arguing that innate knowledge becomes available in phylogeny only because it reliably corresponds to the external world (K. Lorenz, 1981:69); cf. M. Ruse, 1990.

8

Conclusion Whether our knowledge is related to the world depends on the possibility to prove “the relation of that in us which we call 'representation' to the object" (I. Kant, 1980:692). The problem does not concern a posteriori or empirical knowledge, borrowed solely from learnt experience, but synthetic a priori knowledge (SAK), which is about and applies to objects; however, since it is not entirely acquired through sensory individual experience of them, we cannot prove whether SAK actually is causally affected by them. The naturalistic twist to the transcendental aesthetics (science of a priori forms of knowledge, I. Kant, 1980:A22) that came with Evolutionary Theory of Knowledge (ETK), destroyed the concept of a priori; the knowledge of object that precedes its individual experience is a priori in ontogenetic perspective but takes a posteriori character in the phylogenetic one, being acquired through the evolutionary process of natural selection (Lorenz, 1981:101). Given that natural selection instantiates causal relations (E. Sober, 1984), being an adaptation guarantees the relation of correspondence between selective factors and PARs (feature/adaptations, equivalent to the Kantian notion of synthetic a priori knowledge) conferring on the latter the status of knowledge. For ETK natural selection instantiates causality of the same kind as that at work in empirical perceptual experience, as Aristotle understands it, where one and the same form originally present in the external object is transmitted into the organism by a two steps process of evolution by natural selection (variation, selection operating among what is available in the genetic pool). Consequently, ETK implies realism situated in a phylogenetic perspective: if an adaptation exists, the selective factor from the phylogenetic environment responsible for its evolutionary origin must exist too (or at least have existed). The method of reverse engineering is realism-based deduction of selective environmental factors from supposed adaptations, hence a certain kind of circularity. To maintain its realism, Modern Synthesis must be able to 1. unambiguously define and identify traits constituting adaptations. An analysis of adaptation criteria and metaphysical suppositions underpinning them is the aim of the first chapter. 2. prove the existence of a causal, selective, past relation and define and identify unity of selection and of selective cause. The aim of the second chapter is to analyse the means used by the genetics of population to do this. 3. defend its intuitive phylogenetic externalism asserting that selective factors are environmental ones (where environment is considered as independent of the organism). An analysis of the phylogenetic, causal relations environment/organism and organism/environment is the aim of the third chapter. We will show that if PARs (adaptations) exist, it is still far from certain that the Modern Synthesis unambiguously seizes causal phylogenetic relations between PARs and their selective factors and that they externally and environmentally originate (paraphrase of I. Kant, 1980:A 371-372).

I. Two distinct notions: Adaptation and Function We do not expect physical or chemical processes to be useful, as we do in the case of behavioural or morphological (e.g. cryptic) features. Aristotle considered that leaves exist to

9

give rise to fruits; preschool children think that thorns grow so that there is more of roses, while they do not look for a goal of the form of rocks (F. Keil, 1994); in a Darwinian example, the tubular shape of flowers’ corollas is supposed to have evolved according to the length of honeybees’ proboscis. In the light of evolutionary theory, physical, mechanical terms constitute satisfying explanations when applied to abiotic matter but the animated universe acquires its meaningfulness only in the light of the notion of adaptation; without the latter, evolution would be a simple succession of changes (G. Simpson, 1963:106). We read in Darwin’s writings: “At that time (1844) I overlooked one problem of great importance. (…) This problem is the tendency in organics beings descended from the same stock to diverge in character as they become modified. (…) I can remember the very spot in the road (…) when to my joy the solution came to me. (…) The solution, as I believe, is that the modified offspring of all dominant and increasing forms tend to become adapted to many and highly diversified places in the economy of nature” (Darwin’s Autobiography, quoted after G. Simpson, 1963:158-159). In Aristotelian physics or science of nature (in Greek physis means nature) there are two kinds of causes: matter and form; the latter can explain the matter and the why of things, while the converse is not true (Aristotle, Physics: chap.VIII, §6). We find the same in evolutionary biology15: “”How” is the typical question in the physical sciences. (…) It must also always be asked in biology, and the answers can often be put in terms of the physical sciences. That is one kind of scientific explanation, a reductionist one as applied to biological problems: “How is heredity transmitted?” “How muscles contract?” and so on through the whole enormous gamut of modern biophysics and biochemistry. But biology can and must go on from there. Here, “What for?” – the dreadful teleological question – not only is legitimate but also must eventually be asked about every vital phenomenon. In organisms, but not (in the same sense) in any nonliving matter, adaptation does occur. Heredity and muscle contraction do serve functions that are useful to organisms. They are not explained, in this aspect, by such answers to “How?” as that heredity is transmitted by DNA or that energy is released in the Krebs cycle.” G. Simpson, 1963:105– original emphasis). The questions of how are questions about material causes and correspond to immediate causes (E. Mayr, 1961), while questions of why are questions about formal causes and correspond to ultimate (or distal) causes. Just as in Aristotle’s science of nature the fact that stones are heavy and thus attracted toward the earth (material cause) does not explain why stones are the foundation of the walls, in evolutionary biology physical properties of the matter does not explain why living forms are like they are: “If, however, we were asked how apple acquired its various properties, and why it has these properties instead of others, we would need the theory of natural selection, at least by implication. Only thus could we explain why the apple has a waterproof wax outside, and not elsewhere, or why it contains dormant embryos and not something else. We would find that an impressive list of structural details and processes of the apple can be understood as elements of design for an efficient role in the propagation of the tree from which it came. We attribute the origin of selection for effectiveness in this particular role” (G. Williams, 1966:7). Thus, matter alone is not a satisfying explanation, for example of why honeybees’ queens secrete a pheromone preventing workers from laying eggs (M. Wojciechowski and A. Lomnicki, 1987). The theory of kin selection (Hamilton, 1964) explains evolutionary (genetic) benefice of the fact that in the presence of the queen workers stop to reproduct, while in her absence they give rise to male offspring (male because the eggs are not fertilized). The physical reaction for queen’s pheromone, that is explanation by immediate 15

Which shows to what extent “Philosophy is important to biology because biology’s exciting conclusions do not follow from the facts alone. Conversely, biology is important to philosophy because these exciting conclusions really do depend on the biological facts” (K. Sterelny and P. Griffiths, 1999).

10

causation, is not satysfying. Consider another example: during the day fishes better detect zooplankton; this is the evolutionary reason for which the latter migrates to the bottom (Z. Gliwicz, 1986a) and rises to the surface of the water to eat phytoplankton by night. Nevertheless, this vertical migration takes place also each time there is a period of full moon, which is maladaptive for the zooplankton and which is not explained by the influence of an immediate factor such as the moonlight-- the ultimate, distal, evolutionary cause remains valid. Sometimes the influence of immediate, material causes does not explain enough phenomena (like in the case of honeybees’ queen pheromone), or even diverges from the supposed evolutionary, ultimate causation (like in the case of vertical migration of zooplankton in the period of the full moon). Recourse to the latter seems justified. The same goes for functions, apparently close to Aristotle’s form, which is not distorted by accidental, contingent occurrences: the heart’s function is to pump the blood even if particular hearts are sick or damaged and stop to function in the usual manner. Those particular hearts lose some of their properties, without loosing their function. This is the reason why, within EEB, PARs, if taken for functions, can be false, even if their content remains true. This is the case in the situation, where after meaning acquisition period (phylogeny for RAPs or ontogeny for OARs), representations are triggered by a factor, which do not correspond to the object those representations are supposed to refer to. For example, a simple shadow triggers an hiding reaction in goslings, even if it is not actually a bird of prey, but a piece of material used by experimenter. That baby bird has a false belief that x, but does not know that x (there is no knowledge, since the latter requires truth of belief) and we deal with an error: meaning without truth (Fr. Dretske, 1981:195)16. Function and adaptation are considered as something essential: “The biologist who discovers by comparative study that the metabolism and respiratory pigments of animals are closely adjusted to their mode of life is not likely to imagine that the correspondence is fortuitous (J. Huxley, 1963:413-414). Advantages resulting by chance, for example fox’ tracks in the snow around its burrow, are not an adaptation though it spares the energy to reach home next time (G. Williams, 1996:12-13). Thus, in order to find the selective reason for why a trait is what it is we would need to find criteria enabling us to distinguish fortuitous, accessory features from essential ones, that is to distinguish substance or form (Aristotle, Metaphysics: b.delta, §8). “Adaptation does exist and so does purpose in nature, if we define “purpose” as the opposite of randomness, as causal and not a merely accidental relationship between structure and function, without necessarily invoking a conscious purposeful agency. Denial of this does violence to the most elementary principles of rational thought” (G. Simpson, 1963:202). As in Aristotle, there is causal relation between function and structure, given that “natural selection engineers a tight fit” selecting among alternative designs the best solutions to the demands of the environment of Pleistocene (between 10.000 and 1.8 million years). “(1) Information-processing devices are designed to solve problems. (2) They solve problems by virtue of their structure. (3) Hence to explain the structure of a device, you need to know (a) what problem it was designed to solve, and (b) why it was designed to solve that problem and not some other one” (L. Cosmides and J. Tooby, 1994). Task analysis is another versant of the reverse engineering method. Once we know what environmental challenges are, we can more easily find and explain corresponding organismic adaptive responses. For example, in 16

Dretske gives an example of anaerobic marine bacteria endowed with an intern organ called magnetosome thanks to which they go to the bottom where the rate of oxygen is lower and thus environmentally less toxic for them. The bacteria respectively from the northern hemisphere migrate to the geomagnetic north, and those from southern hemisphere to the geomagnetic south, i.e. always far from the surface of water. If bacteria are transported to the hemisphere opposite to the one they originate from, they will migrate to surface of the water and die (Fr. Dretske, 1981).

11

steppe environment all organisms will have similar adaptations (L. Cosmides and J. Tooby, 1994:54-55); in other words, it would be legitimate to infer from the analysis of environmental demands the existence of an adaptation: “Aside from those properties acquired by chance or imposed by engineering constraints, the mind consists of a set of informationprocessing circuits that were designed by natural selection to solve adaptive problems that our hunter-gatherer ancestors faced generation after generation. If we know what these problems were, we can seek mechanisms that are well engineered for solving them” (L. Cosmides and J. Tooby, 1994). For Aristotle the form of a thing can be found in its definition, because the latter expresses its function. For example, the operation of sawing consists on a special manner of dividing; to divide, the tool must have iron tooth (Aristotle, 1993:chap. XIV, §7). Aristotle’s example concerns a tool; reverse engineering consider natural systems as composed of functional units. According to Aristotle’s physics, without the substance or the form, features wouldn’t be intelligible and wouldn’t exist; the matter and proximate causes are separated from the form and ultimate causes. In evolutionary biology the role of the evolutionary cause would play the notion of function: the function of X is Z if Z is an effect caused by X; thus Z explains why X exists, why X evolved in a certain manner (L. Wright, 1973). Spinozian approach Causal relation between two things (where one is considered as an effect and the other as its cause): one can be deduced from the other. The substance is considered as the cause of phenomena. Reason and cause are identified, which gives rise to the possibility of purely rational source of knowledge of reality. As in Aristotle, the why equals the final cause

Functionalist (adaptationist) approach An effect (trait considered as an adaptation) is considered as derivable from an environmental cause. Function (benefice it provides) is considered as the cause of selective origin of trait and taken for an adaptation. Rational engineer’s thought and evolutionary cause are identified, which gives rise to the possibility of rational (reverse engineering) deduction of selective events responsible for origin of adaptations.

In this sense, the notions of function and adaptation are considered as equivalent. Is that justified? The confusion of both notions (adaptation and function) results perhaps from the engineering way of thinking and from too literal an interpretation of Darwin’s analogy between natural and artificial selection. The idea of Darwin was that, given that in limited environmental resources the number of born individuals is greater than the number of individuals that can survive and reproduce, only the fitter ones, even if slightly more than others, will survive. Those originally barely perceptible differences will steadily accumulate and increase until causing qualitative differences and the origin of species (principle of divergence, Ch. Darwin, 1859:ch. IV). Darwin explains the process of natural selection by comparison with artificial selection. Taken to the letter, this explanatory comparison led to some Modern’s Synthesis interpretations of Darwin’s idea, where adaptation is understood as an optimum necessarily resulting from natural selection: “The organisms likely to have more descendants are those whose variations are most advantageous as adaptations to theirs way of life and to their particular environment. Thus evolution is likely to move in the direction of greater or more nearly perfect adaptation, and thus the fact of adaptation, purposeful in aspect but impersonally mechanistic in origin, is explained” (G. Simpson, 1963:196). Elsewhere we find: “Natural selection would produce or maintain adaptation as a matter of definition.

12

Whatever gene is favourably selected is better adapted than its unfavoured alternatives. This is the reliable outcome of such selection, the prevalence of well adapted genes” (G. Williams, 1966:25). If a given feature seems not to be optimal, recourse to constraints is made: “If there were no constraints on what is possible, the best phenotype would live for ever, would be impregnable to predators, would lay eggs at an infinite rate, and so on” (J. Maynard Smith, 1978:32). One of those constraints is trade-off. “(…) any view of biological optimisation that denies the existence of costs and trade-offs is doomed” (R. Dawkins, 1982:47). Here is a metaphorical illustration of this phenomenon:

Selective force drives the wheel (the more baby bird shouts, the more of food it will obtain from its parents) in the opposite direction relatively to the other wheel (the more baby bird shouts to obtain the food, the greater chances it will be a food for its predators). Another kind of evolutionary factors limiting adaptive optimum are historical constraints (S. Wright, 1969): the original form of genotype limits possible directions of its further evolution and reduces chances for a move toward an optimal, global peak in the adaptive landscape; only local peaks are reached. Genetic drift is considered as a possibility of escaping from specialisation (A. Hardy, 1954): the greater the variation, the greater the eliminatory, selective power (R. Dawkins, 1982:20-21). Consider now the well-known, so-called tautology problem of Darwinian logic: what survives is fitter, where fitter (A) means capable to survive and reproduce (B), and survival and reproduction (B) is precisely the measure for fitness (A). This circularity can be eliminated (D. Dennett, 1983) if we take into account that the predicate what survives is fitter cannot be freely exchanged with what is fitter, survives. A variant becomes fitter a posteriori and because it has survived, and not conversely. We obtain then Darwin’s reasoning: S → M [what survives is fitter] Engineer’s reasoning: M → S [what survives, survives because it is fitter] Dennett’s reasoning: ¬ (S → M → M → S) [he predicate what survives is fitter cannot be freely exchanged with what is fitter, survives] If S → M [what survives is fitter], then D. Dennett’s argument is right. However, the question we could ask is whether S→ M [whether survival implies better fitness]. Two points lead us to doubt. First, the notion of adaptation (fitness is a measure for adaptation) is a causal one, supposed to describe real phylogenetic selective relations. Second, evolution is not an engineering process (Fr. Jacob, 1977), since there is neither prior conception nor designer executing the work; natural selection is not an artificial selection process. Consequently there is no necessary equivalence between traits considered as adaptations and features taken for functional units. From a functional point of view we can conceive the features as fitter, less fit, etc. In the quest for adaptations, functional benefit cannot be a cause since it can do not to overlap causal past relations. When Darwin explains how a gradual and non intentional process of evolution can lead to the emergence of complex features, like the eye, worthy of engineer work, he says clearly that what we compare with the telescope, is the eye in its present form (Ch. Darwin, 1859:chap.6). Before performing its present function, the eye was simply a primary tissue

13

endowed with a weak sensibility to the light. During millions of years of evolution, natural selection retained quasi-insignificant, just slightly more light sensitive variations, which subsequently steadily accumulated; what was selected was not the present function of the eye, but precisely immediate factors. Thus, natural selection designates survival of the fittest, or rather survival of the fitter (E. Sober, 1984:176), given that adapted means only slightly better that rival individuals (Ch. Darwin, 1859:ch. IV). The eye was not selected for the function it shares with the telescope, and only what is selected constitutes an adaptation; thus the notion of function does not overlap that of adaptation. Slight differences resulting from immediate and random (in Aristotle sense: non regular, accidental) factors are responsible for selective success, thus instantiating ultimate cause or evolutionary raison d’être. For example, ancestral forms of feathers were probably selected for their thermoregulatory effect, and, from reptilian forms, for gliding and only afterwards for flying; etc. The confusion of the notion of adaptation and function engendered the need to introduce the notion of exaptation. The rain, explains Aristotle, does not fall in order to nourish grains, but because the water first heated convert into vapour, then cooled in the atmosphere condensates and falls in drops; the fact that the rain nourishes grains is a simple accident. Nevertheless, on this accidental factor depend grains’ survival and reproduction; what’s more, material physical properties explain well rain’s influence on the evolution of grains. The same goes to, contingent par excellence, the fire. Whole food chains evolve with reference to the fire, leading to the emergence of specialised cognitive endowments in pyrophiles species (e.g. the antennas of some insects are sensitive to molecules released in the combustion to such an extent that, given that those molecules are different for different trees, those insects are able to distinguish which host is burning; M. Saint-Germain, 2005). The cause of selective success of some plants were birds transporting grains in their alimentary canals to islands more distant than the wind can blow, or honeybees transporting pollen of some flowers. In Greek the same word designates spontaneously and in a fortuitous way, opposed to the word designating both why and final cause. Nevertheless, as the above examples show, the evolutionary why of things can come from accidental causes; adaptation can be identified with some non regular, statistically insignificant factors, that can have nothing or little in common with function allocated by the engineering eye of an anatomist. Functions can be distinguished among all traits of organism (morphological, physiological, behavioural, etc.), which are useful to an immediate goal (like food acquisition, escape from predator, etc.). The functional approach is an anatomical, physiological, mechanical one, using direct observation without having recourse to the phylogenetic, historical origin of traits; today’s utility of trait can or not overlap its evolutionary adaptive origin. On the other hand, evolutionary cause, responsible for the adaptive origin of features can come as well from statistically perceptible as from singular, historical casual factors. An effect of selective importance cannot be automatically identified with a function. Conversely, a function is composed of many effects each contributing to the fitness, but not all should be taken as being of selective origin. Within this perspective, it seems confusing to refer to the term of function (e.g. in Millikan’s notion of direct proper function) where the question is about adaptation (for example in the case of pigmentation of chameleon skin ensuring its survival by hiding it from predators). Originally, teleonomy corresponds to proximal causation and not to the notion of ultimate, evolutionary cause, reserved for the notion of adaptation (C. Pittendrigh, 1958). The difference between immediate and ultimate causes is like the difference between facts and theory: we hope that the former is rooted in the reality, because of their independence from our will dictated by rational thought. The danger with theories is that they arrange the facts, instead of simply sorting them out; the danger of a rational engineer eye is in taking an adaptation (evolutionarily originated feature) for a function, forgetting that the latter does not necessarily overlap with the former.

14

For St. Thomas or Spinoza, the goal is an integral part of the action (omne agens propter finem); thus one can talk about goal only if one deals with an agent (F. Dretkse, 1999). If we considered the world as created, we can conclude that all in nature has a goal. From anaturalist point of view, since there is no agent, natural selection is not an action. This is an obvious reason for which the comparison of natural and artificial selections and their respective outcomes (like the eye and the telescope) has its limits. Function is a rational way of making intelligible the what for of things, but not necessarily the evolutionary why of things, since what engineer’s reasoning conceives and what natural selection produces can or not be the same. Yet, adaptationist deduction is supposed to have not only an heuristic, but an ontological value. The engineering method being approximate is of heuristic, but not necessarily of ontological value. In the next chapter we ask whether statistics, the principal tool of the genetics of populations, are able to satisfy this requirement.

II Genetics of population and its a priori concepts

Taxonomy, or how to define the indefinable 1 7 The variety of species is expected to reflect the different selective pressure’s zones, in which the ecological space is divided. The lack of continuity of the latter implies, within ecological concept of species, a discontinuity in features’ distribution and enables us to objectively grasp the formation of new species (allopatrique speciation). Taxonomy designates classification of species forms and kinships, which “is not arbitrary like the grouping of the stars in constellations” (Ch. Darwin, 1859:chap. XIV) or of classes of objects, but is supposed to carve reality at its natural joints, after the manner of a good carver (Platon, 1922:266) and reflect natural kinds, real units (E. Mayr, 1942). Species classification aims to sort substantial properties (kath’auto- substance) from accidental ones. Until recently taxonomy was metaphysically based on sortal essence (S. Gelman and L. Hirschweld, 1999), i.e. where differentia (i.e. criterion of differentiation in the original context of problem of universals) constituted phenotypic features shared by all specimens. Differentia of being grandmother is not the property of wearing glasses and having grey hair, but of being mother of a parent. For example, ladybirds are not round bugs with red back and black spots, but bugs with 3 tarsal segments; there are grey ladybirds without any spot at all or grey with black spots (Ashy Grey Ladybird); Crioceris duodecimpunctata, species with red back and black spots but with four tarsal segments is classified as Leaf Beetle. Obviously, phenotypic criteria are criticized for being too conventional and are abandoned for causal essence (S. Gelman and L. Hirschweld, 1999), where the differentia constitutes still a property shared by all species instantiations but which is not necessarily an observable feature. Just as visible properties of the water are not causally responsible for its form but the H2O molecule is, genes are causally responsible for phenotypic traits of organisms. Consequently, similitude between species is situated at the level of heritable, genetic properties and differentia constitutes interbreeding community criterion: « Species are groups of interbreeding natural populations that are reproductively isolated from other such groups » (E. Mayr, 1942). Sharing essence means sharing genetic pool and the degree of this sharing can be objectively measured, e.g. between Homo sapiens and Chimpanzee it is 98,4%; 17

Ch. Darwin’s letter from 24th of December 1856 in: Fr. Darwin, 1887:vol.2:88.

15

differentia between Homo sapiens and gorilla is of 2,3%. Thus, the biological species concept is supposed to be free from ‘philosophical’ (here: subjective) element (E. Mayr, 1994/1963). However, sometimes the interbreeding community criterion cannot be applied, in particular in evolutionarily close species, like in the case of African and Asiatic fossils of Homo (J. La Porte, 2005). This particular situation shows the more general problem that classification (here identification of African and Asiatic fossils as Homo erectus, or of African fossils as Homo ergaster and of Asiatic fossils as Homo erectus) depends on the species concept we initially adopt.

Figure 1. J. LaPorte, 2005

If we adopt the biological species concept, the criterion of reproductive isolation cannot be observed; we consider individuals of the initial population (at time t’) as one single species. The cladogram will illustrate phylogenetic relationships between subsequent species as following hierarchy (J. La Porte, 2005):

Figure 2 If, on the contrary, we adopt the phyletic species concept where species is a phylum, a lineage deployed in the continuity of ancestry and descent, “the smallest diagnosable cluster of individual organisms within which there is a parental pattern of ancestry and descent” (J. Cracraft, 1983:170), we will consider individuals of the initial population (at time t’) as two species. The parental branching of the cladistic tree will be drawn in the following manner (J. La Porte, 2005):

16

Figure 3. C is sister species for A and not D, as in the Figure 2, and D is sister species for B.

Figure 4. Two models of speciation (scheme inspired from the one of J. La Porte, 2005): a. b.

cladogenesis, associated with the biological species concept (succeeding typological and phenetic species concepts), where numerical identity is inferred from the qualitative one. anagenesis, associated with the phyletic species concept, where qualitative identity does not imply the numerical one.

The phyletic species concept is combined with an anagenetic model of speciation. Within the initial population (at time t in Figure 4) speciation designates transformation of the whole parental species into a new descendant one - one taxon replaces another without branching; this implies the extinction of the parental species. The appearance of new qualities (new species) occurs within numerically one species, i.e. does not engender the division of the parental population in two distinct species. Thus, species A and B are qualitatively different, and numerically unique. The phyletic species concept is underpinned by the notion of causal essence, where the individuating factor is genetic inheritance; within a lineage genetic continuity of species is kept; thus, numerical identity is kept too, since new qualities (new species) appear without division of ancestral population. The biological species concept, which succeeds the typological and phenetic ones, is underpinned by the notion of sortal essence, naturally matched with the cladogenetic model of speciation. Here, the ancestral population is divided in two or more descendant species but keeps its numerical identity. « If phyletic [anagenetic] transformation (…) does not suffice as a model for the production of new reproductive communities, it follows that the only real mode of speciation is that of splitting, the budding off of a portion of a reproductive to form a new, descendant unit” (N.Eldgredge and J. Cracraft, 1980:21). For example, a group of individuals of a given species separated from the rest by a geographical barrier will form a new species; the parental species keeps its identity and the emergence of the new species (new

17

quality) implies a new numerical identity. In this perspective, the anagenesis is only a special case of cladogenesis, where the appearance of a new species coincides with the extinction of the parental one. The biological species concept is based on the principle of indiscernibility of identicals (Leibniz) asserting that two numerically identical individuals (here: species) are qualitatively indiscernible and share all properties; conversely, two individuals (here: species) that do not share all properties are also numerically distinct18. According to the biological species concept, combined with the cladogenetic concept of speciation, if two species are qualitatively different, there are also numerically different. This is a very intuitive (naïve or folk) metaphysical attitude, abandoned by the phyletic species concept, combined with the anagenetic concept of speciation: here, qualitative difference between species does not imply numerical difference. Another aspect of folk metaphysics implicit in cladogenesis and the biological species concept is based on the principle of permanence: if during a time t there is constancy in perception of new properties, we declare an extrasubjective existence; we do not if during the time t perceived properties vary. The persistence of impressions implies a qualitative identity, from which we deduce numerical identity (the mind-independent existence of object). Nevertheless, this conviction is not acquired by sensory experience, but is inferred by the mind: « (…) the mind is not led to form such a conclusion concerning their constant and regular conjunction, by anything which it knows of their nature. (…) At least, it must be acknowledged, that there is here a consequence drawn by the mind; that there is a certain step taken; a process of thought, and an inference, which wants to be explained. (…) There is required a medium, which may enable the mind to draw such inference (…). What that medium is, I must confess, passes my comprehension; and it is incumbent on those to produce it, who asserts that it really exists (…)” (D. Hume, 1748:partII). The essence (substance) is a converse side of the principle of permanence: we can identify a change (an alteration, a deviation) only by comparison with something else we fix as point of reference: “the permanent, in relation to which alone all time-relations of appearances can be determined, is substance in the [field of] appearance, that is, the real in appearance, and as the substrate of all change remains ever the same”(I. Kant, 1980:B225). Again, substance (essence, objectively existing quality of species) has no empirical origin but constitutes an a priori form of knowledge (a PAR): “if we remove from our empirical concept of any object (…) all properties which experience has taught us, we yet cannot take away that property through which the object is thought as substance or as inhering in a substance (…). Owing, therefore, to the necessity with which this concept of substance forces itself upon us, we have no option save to admit that it has its seat in our faculty of a priori knowledge” (I. Kant, 1980:B6). Thus, substance is characteristic for universal subject: “I find that in all ages, not only philosophers, but even the common understanding, has recognised this permanence as a substratum of all change of appearances, and always assume it to be indubitable” (I. Kant, 1980 :B225). Darwin has the same attitude toward the question: “From the most remote period in the history of the world organic beings have been found to resemble each other in descending degrees, so that they been classed in groups under groups” (Ch. Darwin, 1859:chap.XIV). Recently cognitive anthropology confirmed that essentialism is a part of folk theories, present in every culture examined until now. Essentialism is a conviction that every member of a given category shares a common essence, determining its identity (S. Gelman and L. 18

The principle of indiscernibility of identicals asserts that A=B, if and only if A and B have all properties in common. The converse principle, principle of identity of indiscernibles, asserts that if A and B have all properties in common, then A=B. If A and B are numerically distinct, A and B must have at least one different property. Here, we substitute species for individual.

18

Hirschweld, 1999). For example, essential properties of a particular specimen, e.g. of a particular dog, are shared by every dog. What’s more, people in different cultures, literate or not, classify natural objects in very similar manner, which is called folkbiological taxonomy (or default taxonomy), although orders and kingdoms and not species would be universal (like animals, plants, insects, fishes, birds, mammals, tree, sharks, dogs, etc. (D. Medin and S. Atran, 1999). People universally classify natural systems according to vital interactions they have with them, like edible, nutritious, toxic, dangerous, etc. This rekindles a question as old as philosophy: are genus and species real or are they situated in thought alone (Porphyre, 1998)? The default attitude in the problem of universals is realism (dog is a natural and not only nominal kind) and the value of taxonomy is not only in its pragmatic aspect (pigeon-holing) of naming and sorting out in hierarchies in order to make the variety of species more intelligible and manageable; classification is supposed to have ontological scope. Triggered by sensory ontogenetic experience, but fundamentally innate, folk biology is considered as an adaptation (or PAR). The fact that taxonomical unities constitute adaptations would plead for the position according to which universalia sunt realia ante rem, in other words, species exists before RAPs referring to them. In this sense, the adaptationist argument defending extreme realism has a lot in common with that defending the original radical realist position about Ideas (Plato) and later of typological species concept: if we are bearers of a priori ideas like that of natural kinds, this is because our cognitive endowment have been acquainted with them prior to our births respectively by soul experience during its migration to Pleroma (Plato, 1989:105d19), by God (St. Thomas, C. von Linné) or by natural selection during phylogeny (J. Maynard-Smith). Those habits of the mind would have been retained by natural selection, instantiated by recurrent environmental, selective factors or habits of the world, where world means environment of evolutionary adaptedness (L.Cosmides and J. Tooby , 1994). Is the adaptationist explanation a strong argument for the existence of genus and species and, consequently, for the veracity of PARs? We can argue, as Buffon did, that only individuals exist since we can observe, measure, compare, experience, etc. only individuals, not species: “I can see a horse, but I cannot see horse-ness”. Species are only the names we give while conventionally grouping individuals (a conviction corresponding to the nominalist position, from the latin nominus – name). Nevertheless, at least the mind-dependent existence of those a priori forms of knowledge is undeniable: to conceive a particular horse as a horse, requires a prior notion of what horse is; we do recognise horses as horses, so we do have some concept of horseness. Those a priori concepts exist at least in our minds. PARs, like species, orders, substance, and essence are definitely our epistemic a priori conditions. Can adaptationism prove their mind-independent existence? How can anything prove that our folk ontology transcend to the realm? For the moment, the phyletic species concept, combined with the anagenetic model of speciation, are counterintuitive and go beyond intuitive biological species concept and cladogenesis showing that another way of conceiving the living world is possible.

Units of systematics Thus we surely have an a priori tendency to perceive similarities, enabling us to look for features that a given species does not share with its ancestral forms. Curiously, however: according to our intentions we can see the origin of similar interspecific properties either in common ancestry (the cladistics’ aim to define taxa and to establish phylogenetic 19

Cf. Introduction.

19

relationships between them), or in common selective pressure (the adaptationism’s aim). Indeed, is it supposed that the more similar the species, the less evident their evolutionary relationship: “On my view of characters being of real importance for classification, only in so far as they reveal descent, we can clearly understand why analogical or adaptive character, although of the utmost importance for the welfare of the being, are most valueless to the systematist. For animals, belonging to the most distinct lines of descent, may readily become adapted to similar conditions, and thus assume a close external resemblance; but such resemblances will not reveal, will rather tend to conceal their blood-relationship to their proper line of descent” (Ch. Darwin, 1859:chap.XIII). The greater is the contribution of common ancestry, the lesser the role of natural selection is. This is true especially in the case where the element of common inheritance is excluded, i.e. in the case of homoplasies and synapomorphic properties (similar properties between no related species, called also evolutionary convergences or evolutionary analogies). The more similar, but the less related the species are, the greater role of environmental selective pressure we attribute to their origin, e.g. common selective pressure is the evolutionary cause for white plumage and coats of arctic animals. If we find this equation intuitive and plausible, this is owed to another epistemic a priori condition, namely that of common cause explanation (B. Russell, 1948:216). The latter designates the deduction from two correlated properties of one common cause, rather than of two independent causes. For example, if every bulb in the house blows at the same time, we will find more likely that the fuse has gone (and have recourse to common cause explanation) rather than each bulb has blown independently (and have recourse to separate common explanation, E. Sober, 1984:212). The same goes for homoplasies and synapomorphic properties (not inherited from common ancestor) from which we deduce the contribution of a common environmental cause (E. Sober, 1984:225). Does common ancestry exclude the contribution of common selective pressure in the case of similarities shared by related species, like homologies and symplesiomorphic properties? From a selectionist point of view one can argue that, although from similarities between species one cannot deduce anything about their phylogenetic relationship, the converse works well: the more similar species are (event if related), the greater is the chance that they result from common selective pressure and constitute adaptations. From a non selectionist point of view you can argue with equal rightness, that the equation “the greater the contribution of common ancestry, the lesser the role of natural selection” constitutes an antinomy, since all organisms come from a common ancestor and each organism is related to every other to some degree. Organisms’ properties are what they are in part because they are inherited, and not because they are shaped by natural selection (R. Lewontin, 1978). The alternation between these two explanations (inherited element or adaptation) is not exclusive (either the first or the second), but asymmetrical (E. Sober et St. Orzak, 2003:428). The reason for this is that the common origin is a necessary but not sufficient condition for a property to be shared in succession by species within a lineage; a continuous control of stabilizing natural selection is still required. For example, if predators and with them selective pressure disappear, mimetic ocella on the wings of Papilio dardanus, resembling the eyes of insectivorous predators of butterfly’s predator, disappear too (E. Mayr, 1994/1963:165-166). Thus, (asymmetric) exclusivity results from the fact that to be satisfying, common origin explanation of similarities in homologies and symplesiomorphic properties (properties shared by related species) needs to be accompanied with their selective raison d’être. A versant of the common cause explanation is causal explanation tout court: from a property (effect) we deduce contribution of its cause, rather than the lack of causal origin. Compare with Spinoza: every effect has its cause. Thus, the difference in shell patterns of Cerion’s specimens respectively from Long Island and Acklins will have its satisfying explanation in

20

the functional, survival value of the pattern (E. Mayr, 1994/1963:269), rather than in the difference between the histories of snails (St.J. Gould and R. Lewontin, 1984/1979).

Natural selection and/or drift The attitude ascribing organisms’ forms to an evolutionary cause is a deterministic one. From this point of view, hazard means temporary ignorance of the cause: « The exact proportions of the various species of plants, the numbers of each kind of insect or of bird, the peculiarities of more or less exposure to sunshine or to wind at certain critical epochs, and other slight differences which to us are absolutely immaterial and unrecognisable, may be of the highest significance to these humble creatures, and be quite sufficient to require some slight adjustments of size, form, or colour, which natural selection will bring about » (A. Wallace, 1899:148). To consider a phenomenon as random, means to acknowledge “our ignorance of the cause of each particular variation” (Ch. Darwin, 1859:chap.V,§I). From this point of view, genetic drift governing the evolution of neutral mutations is considered not as an evolutionary force alternative to that of natural selection, but rather as an aspect of the latter: stochastic phenomena affect samplings exposed to natural selection, but do not superimpose on the latter. Thus, even if a whale swallows half of a plankton population (random sampling), the probability of reproductive success of 50% or 0,001% of the population left is determined by natural selection (discriminating sampling) (E. Mayr, 1994/1963:123). From an omniscient point of view, drift or hazard instances would be eliminated (A. Rosenberg, 2001). We can have an indeterminist attitude toward natural objects and consider hazard as the absence of the cause. In the first position we assert the lack of the knowledge of the cause (epistemic scope), in the second we assert the lack of the cause (ontological scope). Cause is instantiated by natural selection, and hazard by the notion of drift20. Intuitively, the latter is considered as distinct process of indiscriminating sampling, alternative to that of natural selection (discriminating sampling). Even if we were in possession of the complete knowledge of all phenomena, we would still have to accept that part of evolution is due to random drift, for example if a forest fire kills more of grey squirrels than of the red ones, the sampling of genotypes determining grey coats from the genetic pool of that population is random- grey squirrels are not dead because of the colour of their coats (R. Millstein, 1996:S16). However, though the adaptationist, selective and causal explanation, e.g. of the why of Cerion’s shell patterns, is more satisfying than the one which does not look for their selective meaning; neither the first, nor the second explanation, is proved.

The power of natural selection What then is the actual contribution of natural selection to the evolution of natural systems? If natural selection were omnipresent, every new mutation should be selected, either positively (replacing present allele), or negatively (lethal genes are eliminated from the genetic pool). In consequence, polymorphism should be lost. However, natural population are 20

Genetic drift explains random fixation of alleles in small finite populations (e.g. fixation of alleles whose value is smaller than that of other alleles existing in population). Founder effect shows that change in genes’ frequency and a genetic differentiation between two populations can take place rapidly and in a random, non adaptive way, e.g. in a situation when a small part of a population is separated (e.g. by a geographical event, like change of a river’s flow); the individuals of the latter do not constitute a representative sample of genetic variants present in the original population; by chance a number of alleles will be lost and not represented in the new colony.

21

very heterogeneous at the molecular level (for each gene there exist many alleles) as well as at the phenotypic one (for each phenotypic propriety there exist many alternative morphs). One reason explaining that state of matter is the selective neutrality of the majority of mutations, in particular synonymous mutations (one amino acid is replaced by another having similar properties) and silent mutations (in spite of an alteration of one of three consecutive nucleotides- in 70% of cases the 3rd one- codon specifies the same protein and its function remains the same- no selectively significant effect) (M. Kimura, 1983). Since about 10 millions years are needed (i.e. average life time of one species) in order that one mutation reaches sufficient frequency to be visible for natural selection (R. Lewontin, 2003:108), it is clear that to some extent the evolution of neutral mutations takes place by genetic drift. The fact that polymorphism exists can be seen in a twofold way, that is not as a consequence of selective neutrality and thus as limits to the power of natural selection’s control, or, on the contrary, as a consequence of adaptiveness of heterogeneity (T. Dobzhansky, 1968). The existence of variability would be maintained by natural selection itself. Given that the change of gene frequency is proportional to the variation of that gene, the greater variation, the greater possibility of rapid (and adaptive) changes. Thus polymorphism would confer an evolutionary advantage in a changing environment. Natural selection acts on the phenotypes and neutral mutations affect only molecular level. This is seen not as concurrent to natural selection action, reducing its power, but as complementary (R. Dawkins, 1982). However, if natural selection acts on phenotypes, only adaptive traits will be selected, whether underpinned by one single or many genes. The power of natural selection control can be seen as weakened, since very often many alleles, and not one, are responsible for a property (that is in the case of polygeny, when many nonallelic genes control the expression of a trait), the intensity of the action of natural selection on each locus may be weaker (St. J. Gould and R. Lewontin, 1984/1979:82). The converse situation of a pleiotropy can also provide the possibility that some properties, which are one of many effects of pleiotropic gene21 can enter the next generation thanks to other phenotypic effects the gene has. For example, if a selectively harmful or neutral propriety X is linked by association to an advantageous property Y, X will be transmitted to the next generation, but we will not be allowed to say that X was selected and constitutes an adaptation (E. Sober, 1984:196-197)22. Such associations are engendered by phenomena such as genetic linkage (association of genes on the same chromosome, that causes them to be inherited together), epistasis (one nonallelic gene determines whether other genes are expressed or not), pleiotropy, polygeny, heterozygosity, etc. Consider first an example of such associations at the molecular level: when two given alleles are independent, their respective results are selectively harmful, but when those two alleles are matched, they conceal their respective disadvantageous effects; the association of those two alleles has neutral selective value. The time needed for such an association to be fixed in a population is relatively short, though 10 time longer than the time needed for a single neutral mutation to be fixed (J. Szwejkowski, 1987:390). Another case is heterozygosity. Mutations can slip into the genetic pool of the next generation in spite of their harmfulness when in recessive condition. Of course, if a disadvantageous mutation occurs on the gene in homozygous condition (i.e. in diploid

21

Called also polyphenic, gene determining many traits, e.g. skin colour and size in mouse. A similar problem concerns functional considerations: which property was selected for and constitutes an adaptation and which one is only a by-product or spandrel, e.g. the mobile hedgehog-like spines of sea urchin were selected for locomotion, or maybe for food transport. 22

22

organism mother’s and father’s alleles are identical23), it causes the death of its bearer and involves negative selection- no transmission to the next generation. However, if a disadvantageous mutation occurs on the gene in heterozygous condition, its effect disadvantageous trait-, is not expressed; the mutation is selectively neutral and enters the genetic pool of the next generation24. The important conclusion is that, under condition that the gene is completely recessive, it can be kept in a population irrespective of its effects, thus remaining beyond selective control. On the other hand, however, this quasi independence of associations is seen as enabling adaptive evolution (R. Dawkins, 1982; J. Bonner, 1988) and the evolution or modules, or PARs. Quasi-independence makes natural selection more efficient since through the selective pressure aiming at phenotypic targets-traits, the genetic basis of the latter are affected (R. Brandon, 1999). Such relatively fixed and hardly divisible by natural selection relations are basis for modules, a notion precisely designating trait/gene match; modules are identified through their specialised functions. In short, modules have mostly inner pleiotropy, that is traits/genes matches are weakly connected to other traits/genes (R. Brandon, 1999). “Quasi-independence means that there is a great variety of alternative paths by which a given characteristic may change, so that some of them will allow selection to act on the characteristic without altering other characteristics of the organism in a countervailing fashion” (R. Lewontin, 1978:230). The existence of such units is a basis for engineering and systematic analysis. For example, mammals’ front paws constitute such a functional inherited unit (R. Brandon, 1999:174) that could be identified in cladistics as homology or symplesiomorphic property (adaptive features shared by related species).

Possibility of modularity is based on the existence of the interdependence between sets of genes [here two: (G1, G2, G3) and (G4, G5, G6)] determining expression of phenotypic traits [here: C1 = (A, B, C, D) and C2 = (E, F, G)] accomplishing different functions [here respectively two functions: F1 et F2] (G. Wagner, and L. Altenberg, 1996:97l).

Units/modules/PARs result from genetic linkage, pleiotropy, recessivity, epistasis, etc., trait/gene matches sharing the same function and commonly inherited. There exist the 23

Note that in haploid organisms disadvantageous genes are immediately eliminated, since there is no allele to mask their harmful effect. 24 This is the reason why hereditary diseases are mostly recessive and transmitted by woman lineage (women have two X chromosomes, a disadvantageous mutation on Xa will be hided by the second Xb. In men (XY) because of chromosome Y, each disadvantageous mutation occurring on X chromosome have the effect of dominant gene (St. Cebrat, 1998). Recessive means of little or of no effect when accompanied with a contrasting allele, expressed only in pair with identical allele, i.e. in homozygous condition. Dominant means expressed also in heterozygous condition, i.e. in pair with recessive allele.

23

possibility that selectively harmful or neutral proprieties are in, entered thanks to the association with other advantageous properties of the unit/module/PAR. Nevertheless, we are not allowed to say that such free riders were selected and constitute adaptations (E. Sober, 1984:196-197). The question is, precisely, whether such free riders get to the next generation by chance, or their entering is under natural selection’s control and, if the first, whether modules/PARs are composed of non-selected and non-adaptive elements. To answer that question we need to know what level natural selection aims at.

The problem of the level of selection Consider two examples. The first one concerns population of moths with stripes on the wings resembling grooves in tree bark, that morphologic trait being determined at a single genetic locus. Because some moths have transverse stripes and other longitudinal ones, moths are camouflaged only when sitting in the right position towards the bark’s grooves, respectively horizontally and vertically, that behavioural trait being determined at a second single genetic locus. What ensures camouflage function and thus high fitness value is the association of those two traits (i.e. transverse stripes/sitting horizontally and conversely) and genes underpinning them. The second example is that of superiority of heterozygosity: individuals with homozygous genotype aa die from sickle-cell anaemia inhibiting the production of normal red blood cells. In individuals with homozygous genotype AA sickle-cell anaemia is not expressed, but they die from malaria. In individuals with heterozygous genotype Aa or aA, although they bear recessive genes of sickle-cell anaemia, the disease is not completely expressed (and not mortal) and in addition, modifications of red blood cells have the advantage to defeat malaria. In short, heterozygous individuals do not die either from sickle-cell anaemia, or malaria. The question is: where does natural selection occur then, at the level of single gene, or at the superior level of two alleles’ association? In the first example, properties, respectively of transverse stripes and of sitting horizontally, if taken alone do not protect moths from being seen and eaten. Fitness depends on the association of both genes/traits. Thus, the property of having transverse stripes is fit in one context, i.e. when accompanied by the property of sitting horizontally; the fitness of the property of having transverse stripes alone is inferior. In the second example, if we take the single gene as the level of natural selection action, allele a increases the fitness when accompanied by contrasting allele A on homologous chromosome, but decreases the fitness when accompanied by identical allele a. The problem is that from the gene perspective there is no selection for heterozygous genotype Aa; yet there is selection for homozygous genotype aa—its bearers are dead because of sickle-cell anaemia. sickle-cell anaemia

malaria

selection if: single gene level

selective advantage

homozygous: AA

no

yes

homozygous: aa

yes

death anyway

yes

yes

heterozygous: Aa

little

no

no

only if malaria in the environment

24

Superiority of heterozygosity. Gene’s point of view.

However, the problem with the gene point of view is that it does not identify the real causal events, if ‘real cause’ cannot be the cause in one context, and not be the cause in another (E. Sober and R. Lewontin, 1982; E. Sober, 1984; W. Wimsatt, 1991). For example, in the case of superiority of heterozygous, selective cause cannot at the same time increase and decrease the fitness, or it is not the real selective cause (E. Sober, 1987); thus, gene, from this point of view, cannot be seen as the selective cause. In order to find the real, contextindependent cause for both examples we should look for it at a higher level: “if the fitness of x is context sensitive, then there is not selection for x; rather, there is selection at the level of organisation higher than x”. (E. Sober, and R. Lewontin, 1982:169). In the case of sickle-cell anaemia, selective coefficient should be calculated not according to gene alone, but according to diploid genotype, given that not gene alone contributes to the fitness but the association of two alleles (Aa or aA). Can this realistically motivated quest for the cause be fulfilled? If we wanted to apply the criterion of context independence consistently, it would not be fulfilled even by diploid genotype, since the heterozygous genotype is well fit in the context where malaria is present in the environment, but is inferior in the context when there is no malaria in the environment; the fitness value of the diploid genotype is context dependent too (J. Maynard-Smith, 1987). In practise, context independence criterion of recognition of level of selection can be fulfilled if it takes a quite different form, compatible with the gene level (K. Waters, 1991). As an example, consider the situation where a lethal dominant gene can be suppressed by an allele. Among two separated genes populations, there is one (I) where all individuals are homozygous for a repressor (gene inhibiting the expression of another gene), and there is another (II) without repressor. Within (I) neither positive no negative selection of lethal gene occurs, because there is no repressor, while within (II) lethal gene is eliminated (negative selection); the fitness of the gene is context-dependent. Now, according to the original context-independence criterion, the action of natural selection should be identified at a higher level than the gene alone25. Practically, however, within population without repressor (II), negative selection against lethal gene is inferred, while within (I) no selection action is declared. The criterion dictated in practice has this form: “if the fitness of x is context dependent, then either (1) selection is acting upon a level of organisation higher than x; or (2) there are several different selection processes occurring at the level of x” (K. Waters, 1991:558). That criterion is compatible with gene selectionism: in the case of homozygous genotype aa (individuals die because of sickle-cell anaemia) there is selection at the level of gene alone, while in the case of heterozygous genotypes Aa resistant for malaria selection it exerts its action at the level of the environment of the gene, i.e. its homologous allele A. This case shows that natural selection is a frequency-dependent phenomenon, that is selection of a trait is determined by its frequency. For example, if sex equilibrium leans toward males, female births will be favoured (and conversely); thus selective value of the property of having female descent depends on its frequency. The same goes for the case of superiority of heterozygosity: in the environment where malaria is present, allele of sickle-cell anaemia is favoured when not frequent, i.e. accompanied by normal allele and eliminated when frequent, i.e. accompanied by identical allele (K. Sterelny and P. Kitcher, 1988). The fact that the gene alone is not sufficient for a trait to be expressed does not imply that the gene alone is not the level at which natural selection acts: the unity of development 25

Consistently with their argument that if the fitness of x is context-dependent, the level of selection is not x, but a higher level (E. Sober, and R. Lewontin, 1982:169).

25

(regrouping all factors necessary for a trait to be developed) is a different question than that of unit of selection (or unity of inheritance). Development does not determine what is inherited (R. Dawkins, 2004). In the example of moths, morphological genes (longitudinal or transverse stripes) need to be combined with behavioural ones (sitting horizontally or vertically) for the whole phenotype to be camouflaged and thus maximally fit. It does not imply, however, that those two properties are affected on the whole. During meiotic process of crossing-over genes are interchanged and randomly recombined and new assortments are formed. This gives a possibility for natural selection to eliminate less efficient genes from previous associations (R. Dawkins, 1982/1999:241). The reason for this is that entities that are not heritable cannot be adaptations of those entities (J. Maynard-Smith, 1987), and that, although genes are selected for theirs phenotypic properties, only genes are inherited (R. Dawkins, 1982). Thus, if within a module (PAR, like that of camouflage in moths) one phenotypic effect is harmful and another beneficial, there is no reason that natural selection will not separate those effects and genes underpinning them (R. Dawkins, 1982:35). Indeed, gene selectionism takes for (selective) cause what holism takes for effect (E. Sober and R. Lewontin, 1982). However, for both camps only the gene is inherited, and for both the reason why the gene is inherited is the fitness of the phenotypic result of the association of genes. The holists’ point: the fitness of diploid genotype instantiates cause while fitness of a gene instantiates effect is compatible with the gene selectionists’ point: through the fitness value of genes associations (diploid genotype) single genes are selected, inherited. The core of the quarrel concerns rather the realistic expectations of holists, who want to find the level of selection, the cause. If sometimes x is the cause and sometimes is not, this identification fails. Gene selectionism, in contrast, have instrumentalist pretensions, where selection can be situated at the level most appropriate to the experimenter’s aims, e.g. at the level of macro-population if we analyse the dynamics of a population, at the level of micro-population if the study concerns individuals, etc. (A. Ariew, 1998). Statistics provide us with the knowledge of correlations and variations between statistical properties of classes of traits, and with calculations of a probable frequency of traits (D. Walsh, 2004). This is precisely problematic for realists, for whom population genetics do not identify selective causes, but only link one set of effects to another, those sets being abstract and purely numerical properties, without identifying particular selective causal events, responsible for the reproductive success of particular individuals. The cause, in the realist sense of the term, is grasped only by token causality (E. Sober, 1987) or causal singular assertions (St. Glennan, 2002), referring to single, particular events (e.g. my cousin, a non-smoker, died of lung cancer by reason of x). In contrast, property causality (E. Sober, 1987), or causal general assertions (St. Glennan, 2002) of population genetics, referring to causal role of some properties in some populations (e.g. smoking causes lung cancer), would not have any ontological value given that population’s property is not a property of particular individuals from that population. For example, the average income of a family of four is the total income for all families of four divided by the number of those families. Although the average income depends on the incomes of all families, the income of a particular family does not necessarily correspond to the average income (B. Horan, 1994:79-80). From aninstrumentalist point of view, probability also concerns individuals, given that the 50% probability of a coin to fall on heads is as well valid for that particular coin as for a set of coins (A. Ariew, 1998). Property causality are legitimate causal assertions (J. MaynardSmith, 1987). To consider A-s as the cause of B-s, not every A should be followed by a B, and not every B should precede an A. If it happens that it is not the case (e.g. my cousin died of lung cancer and he was a no-smoker), A is still the cause of B in the majority of cases. The premise that A causes B (that smoking causes lung cancer) enables us to make better statistical predictions about B-s than its absence. Thus, assertion that allele a increases the

26

fitness when with contrasting allele A on homologous chromosome, but decreases the fitness when with identical allele a, is a legitimate causal proposition. Can we really rely on probabilistic and statistical analysis, where gene is the cause? For example, the probability that a disease will be expressed is greater in an individual bearing gene x than in another whose genotype lack the gene x, e.g. probability that mongolism will develop is greater in individual bearing a supernumerary chromosome, than in an bearing diploid chromosome number. However, genes alone are not sufficient for reliable predictions, because they are not sufficient for a trait to develop, for example in the case of phenylketonuria we can prevent mental deficiency by introducing an appropriate diet (L. Gannett, 1999). If environmental factors are equally needed for a property to develop, we cannot attribute the status of the cause to gene alone. In principle, many factors causally contribute to the selective value of the phenotype, some more directly than others, some coming from molecular level, and others from the visible phenotypic one. It implies, that if we exclude the potential causal influence of some of them, we conceive causes in a hierarchic way. It would be essentialist to consider genes’ causal contribution as self-sufficient and genes’ environment causal contribution as accidental. This essentialism rests on the very intuitive consideration that even if gene is not the only cause, gene is the cause, just like striking a match is the cause it lights up, i.e. is the necessary and sufficient condition, while background conditions (such as an appropriate oxygen rate in the atmosphere, etc.) are just necessary (K. Sterelny and P. Kitcher, 1988). Is that justified? Background conditions are normal conditions. The same goes for the validity of assertion: Down’s syndrome has genetic causes, which is based on the fact that disease develops in the normal context. Of course, assertions of this kind are true only for every known context and to be the cause of x in every known context (epistemic scope) is not the same as to be the cause of x in every context (ontological scope) (L. Gannett, 1999). It does not reduce the instrumental value of this assertion, and there is another way to justify this kind of causal essentialism. From an adaptationist point of view, causal general assertions are reliable not only because are based on particular cases, but also because natural phenomena occur with sufficient regularity: natural systems have a capacity for self-regulation. Even if deviations of normality can of course occur, the causes of their dysfunction are independent of organisms, e.g. driven by natural disasters. Normal states must statistically prevail over dysfunctions, otherwise natural systems would not have survived and would have been eliminated by natural selection; positive selection for a trait must have led to its statistical predominance and every dysfunction must be statistically rare: ”why are self-regulatory systems omnipresent in our world? Where do theirs prototypical norm states come from? Why do their self-regulatory mechanisms normally work properly? Our answer is: because of Evolution, in the generalized ‘Darwinian’ sense of natural (…) selection” (G. Schurz, 2001:480). Thus causal general assertions in general are true, although not in all contexts: “prototypical normality and statistical normality are connected by the law of evolutionary selection” (G. Schurz, 2001:480). Once again we find the petitio principii, given that reliability of causal general propositions supposed to prove natural selection’s action, would be founded in regularity, and regularity is explained by natural selection26. Another circularity would concern the notion of selective cause itself. The latter designates the difference between genotypes’ fitness; 26

The same reproach of circularity goes for functionalist analysis. First we explain the adaptation as necessarily resulting from the process of natural selection, to support the notion of function: dysfunction must be statistically rare, otherwise organisms would not have survived and would have been eliminated by natural selection; Then we circumscribe normality in order to define function; yet you cannot circumscribe regularity without having defined previously the function. The notion of normality itself is a normative notion.

27

however, the latter is defined and determined by its effects, i.e. quantified by the number of alleles passing to the next generation; the way of grasping selective causes is circular (B. Horan, 1994:88).

Conclusion Even if an omniscient point of view grasping all causal relations is not within reach, it does not imply that all or no explanations are good and if we cannot have a complete knowledge it does not imply the rejection of the partial knowledge we have: without being totally good, context-dependent, or frequency dependent causal assertions, can still be quite good if we decide to temper realist pretensions and to concede “that population-level forces can increase the chances of theirs effects in some contexts and decrease in others” (K. Waters, Kenneth, 1991:555). That kind of inconsistencies or situations where facing the same data we can as well reject as accept assertions enables us to go beyond our a priori concepts toward a metalevel of conceiving reality. This way, genetics of population enables us to see, for example, that neither natural selection, nor genetic drift can by considered as alternatives forces (at least we cannot prove their existence in themselves). If the omniscient Laplace’s demon would see the trees, he would not see the forest (E. Sober, 1984 :126-129); the same goes for genetics of population: if we had the possibility of conceiving simultaneously all singularity and variability of particular phenomena the notion of evolutionary force would be cleared away. Notions such as (selective) cause or natural selection, drift, regularity, biological essence and species, unit of systematic (unit of inheritance), function and functional unit, etc. are all to some extent epistemic a priori concepts. It is still far from certain that those concepts reflect mind-independent phenomena. What is certain, is that there is another consistent way of conceiving reality, at least as efficient as that dictated by a priori intuitive concepts. If a priori concepts are adaptations, we could believe that because the degree of this adaptation varies from one species to another, PARs give different (as different as species) but never contradictory images of the world. In the next chapter we will see that the only ‘objective’ criterion of veracity is that of viability, and that criterion does not imply the externalist view of adaptation, according to which living organisms adapt to one and the same world, where world means selective environment! The last but not least question is whether natural selection has a control on free 27 riders . In principle, lethal mutations can enter genetic pool if mutation occurs on the gene in heterozygous condition, that is the gene is not expressed. Lethal mutation could not slip to the next generation if gene is in homozygous condition, because it causes the death of its bearer. As it have been said, if indeed within a module (PAR, like that of camouflage in moths) one phenotypic effect is harmful and other beneficial, there is no reason natural selection will not separate those effects and the genes underpinning them (R. Dawkins, 1982:35). However, homozygous condition does not guarantee negative selection and genetic death of disadvantageous gene! The reason is that housekeeping genes switch on chronologically, and some of them after reproductive age of organisms. Thus, if a disadvantageous mutation occurs on such a lethal gene, its selective value is neutral and it is transmitted to the next generation (St. Cebrat, 1998). The more genes switch on late in the ontogenesis, the greater chance they are disadvantageous, or that they are genes of death responsible for senescence illnesses, like Alzheimer, Parkinson, etc. (St. Cebrat, 1998). Since selective value does not quantify how 27

What we previously called component of units/module/PARs, which are selectively harmful or neutral proprieties, entered to the genetic pool thanks to their association with other advantageous properties of the unit/module/PAR (associations resulting from genetic linkage, pleiotropy, recessivity, epistasis, etc., i.e. sharing the same function and commonly inherited).

28

many years we have lived, but how many descendants we had, even if natural selection were omnipotent, phenotypes would not live forever.

III. Relativist Theory of Evolution Causal phylogenetic externalism of Modern Synthesis In metaphysical dualism, on which modern theory of knowledge was based (since Descartes), sensory experience can be determined only by a (sensory) part of representation, of how an object is given to the subject. This results from the supposition that body and mind are two different substances: divine cause had the power of determining soul’s (mind’s) states, which did not come from the external world, material causes. Metaphysical dualism is a consequence of the asymmetric manner of conceiving causality. The first evolutionary theory of modernity (Lamarck, 1809) is also underpinned by an asymmetric conception of causality: environmental changes trigger evolutionary process by modifying survival demands organisms face and the environment somehow induces variation. Some organs are used more often than others. For example, because of reiterated effort to crawl, snakes would have acquired their tapering form and inversely would have lost their paws as a consequence of not using them. The second important moment of the evolutionary theory of Lamarck was inheritance of acquired traits. This theory states that the parental organisms transmit to their offspring the traits that they acquired in ontogenesis. Obviously, this point of the Lamarckian hypothesis has never been proved. In Modern Synthesis, variation originates spontaneously and randomly, where acquired traits do not affect an organism’s genome and only genome (and not what parents learned or acquired during their ontogenesis) is passed to the offspring28. Nevertheless, Modern Synthesis, though underpinned by metaphysical monism (materialism), shares causal one-sidedness with modern theory of knowledge: “we are products of processes independent of ourselves – to suppose otherwise would be to make ourselves God. Our ancestors and now we (as it were) bounce(d) off against reality and thus they/we refine(d) our sense organs and powers of thought. The reason why we see (say) a chain is because it exists – independently of us – and it is in our interest to know of this existence” (M. Ruse, 1990:105). There exists two causally independent entities: ecological niche and organism; only one of them (environment) has selective evolutionary influence over the other (organism). Obviously, the theorists of evolution are conscious of the mutual influence between the environment and the organism: “No animal, no plant can live in the vacuum. A living organism exchanges continuously substances with its milieu. The tree absorbs water and mineral elements by its roots while it lacks of water and it absorbs carbon monoxide by its leaves. A mammal absorbs water and nutritive substances in its intestine and oxygen in its lungs. Without these exchanges the life is impossible” (J. Maynard-Smith, 1962). However, they continue to consider adaptation as a unilateral process where the organism adjusts itself to the environment, which is fixed in a previous context. John Maynard Smith continues: “The life is thus an active balance between living organism living and its milieu, balance which can

28

In 1904 August Weismann showed that the germ line is segregated from the soma, thanks to the observation that the offspring of mice with cut-off tails has normal tail. Since 1958 (after discoveries in molecular biology of Watson and Crick) Central Dogma of molecular biology sets out that DNA causes the production of RNA that makes proteins and then cells; the reverse process doesn’t occur.

29

be maintained only if the surrounding milieu is appropriate for this plant or this particular animal, that is considered then as “adapted” to this milieu”. According to Modern Synthesis (MS), adaptation is a “mechanism thanks to which external cause is transformed into effect”(R. Lewontin, 2003:118-120), an asymmetrical process where “the environment brings about an organic change exactly in its own image” (P. Godfrey-Smith, 1996:86), and where “organisms adapt to their environments, never vice versa” (G. Williams, 1992:484). In MS, the movement of natural selection is environmentally driven (the environment differentiates between two genotypes G1 and G2). In this externalist concept of the phylogenetic relation environment/organism, adaptation designates population’s moves relative to selective environments (R. Brandon, 1990:45), a set of factors differentiating reproductive rate of individuals29, also referred to as mesocosme (G. Vollmer, 1984), micro habitat or ecological niche. These are terms designating those (set of) factor(s) that have a selective importance for the evolution of a given species. The survival of the fittest among all genotypes in the population designates the fact that genotypes are selected (i.e. reproduce and pass their genes to the next generation) according to their environmentally determined fitness function. The fitness (adaptive value designating the probability of survival and the rate of reproduction of a genotype) constitutes the measure of the efficiency with which organisms track the form of essential characteristics of the external world. The organisms are considered as “passive objects moulded by the external force of the natural selection” (R. Lewontin, 1983:275). The form of the environment is printed in the form of an organism and adaptation designates the gradual phylogenetic environmental moulding of the organism (P. Godfrey-Smith, 1996). The Direction of the evolutionary movement depends only on the environmental selective pressure. The organism seems to be too small and the world too big (J. Odling-Smee, 2003:31). Compared to the environment, the organism is not able to modify or construct its selective environment to an evolutionarily significant extent. The organism did not create the stars or the mountains30! The formal model of evolution by natural selection as the Modern Synthesis understands it, is the following one (R. Lewontin, 1985): i. ii.

dO/dt = f(O,E) dE/dt = g(E)

Equation (i.) formally represents the phylogenetic and adaptive change in the organism, driven by two variables: the environment and the organism. Equation (ii.) formally represents the change in the environment driven only by an environmental variable.

Within the EEB, PARs constitute “picture of the environment”, which reflects its permanent and repetitive properties, “data from the external world, which are present with sufficient statistical regularity over long durations” (K. Lorenz, 1975:64). Phylogenetic analysis of the milieu allows discovering to which factors PARs constitute an adaptation (J. Tooby and L. Cosmides, 1989). Given that natural selection is identified with (a set of) environmental factors that influence the survival and reproduction of the organism and given that natural selection instantiates causal relations, an adaptation guarantees the relation of correspondence between selective (environmental) factors and PARs (feature/adaptations,

29

According to this distinction (R. Brandon, 1990), natural selection operates within selective environment; ecological environment includes those factors that influence survival and reproduction in an indirect manner; external environment designates the rest of factors that have no causal effect on evolution of organisms. 30 Paraphrase of:: “Human minds did not create the stars or the mountains, but this “flat” remark is hardly enough to settle the philosophical question of realism versus antirealism” H. Putnam, 1992:30.

30

equivalent to the Kantian notion of synthetic a priori knowledge) conferring on the latter the status of knowledge.

The organism constructs certain aspects of its world Organisms eat to live, kill to eat, and evacuate because of eating, etc. - living organisms consume and transform the matter and the energy, changing chemical composition of the environment by secreting chemical substances resulting from metabolic processes, by using the energy, by exploiting the resources that a milieu offers (e.g. lions predating antelopes will have an influence on their number), etc. The photosynthesis of at least 260,000 species of plants, bacteria and algae, changed the composition of the atmosphere (R. Lewontin, 2003). The vegetable and animal plankton contribute to the exchanges of the oxygen and of carbon dioxide between subtropical oceans and atmosphere (R. Day et al. 2003). Finally, forests regulate hydraulic balance to the extent of determining the weather. The organism actively changes certain biotic or abiotic aspects of the environment31 (the phenomenon bears the name of perturbatory niche construction) by building artefacts (birds, fishes, ants, termites, etc, build their nests, the bees their hives, butterflies and spiders their cocoons, many mammals their burrows, beavers their dams, etc.). The organisms choose theirs microhabitats (this phenomenon is called relocatory niche construction), e.g. migratory birds; gazelles and zebras cross the plains of the Africa following the rain; the oceanic plankton moves according to a daily rhythm, etc. Though organisms did not create the stars or the mountains, there are many aspects of the world that are literally created by them. In certain evolutionary situations, individuals of a given species create the properties of ecological and selective environment of other species. For example, the excrements of ruminant herbivores constitute food for the beetles. The survival of beetles depends on those nutritive resources, i.e. the presence of the dung. In other evolutionary situations, one species evolves relative to another. For example, the form and the smell of Cryptosilis is similar to the female of the fly which, tricked, tries to copulate with the flower, and thus ensure the transport of the pollen from stamens to pistils. What’s more, “organisms not only determine, by their form or by their metabolism, the aspects of the outside world which are important for them, but they build actively, in the literal sense of the word, a world around them” (R. Lewontin, 2003:66). Consider a property one would intuitively describe as intrinsic to the world: the oxygen- rich atmosphere of the Earth. A majority of organisms cannot live in its absence. It is an environmental factor generating “a need” and “imposes” constraints on ‘adaptations/responses’ such as ensuring oxygen’s distribution circulation. Nevertheless, high oxygen rate, necessary for the majority of living creatures currently living on the Earth was much lower at the origin of life. Furthermore, oxygen was toxic for the organisms living at that time. Once, at an evolutionary moment, metabolic reaction between a modified cell and the oxygen took place. Gradually, the proportion of the components of the Earth's atmosphere was reversed and the anaerobic organisms constitute a minority today. High rate of oxygen in terrestrial atmosphere emerged with a change of metabolism of the living organisms, subsequently becoming a universal constraint to which the evolution of the majority of living creatures was subjected.

Relativist Theory of Evolution Thus, it is an empirical question to know whether the organism changes the ecological and selective environments relative to which it evolves. We can dispute which aspect of the 31

Examples from R. Lewontin, 2003 and J. Odling-Smee, et al, 2003.

31

world can be included in the set of properties of the ecological environment literally created by organisms, but more essential is the fact that organisms changes their selective environment. Given that the scope of natural selection is local, natural selection being known to be short sighted, transformations brought by the organism to the environment can have a considerable effect and modify local selective pressures (J. Odling-Smee et al. 2003:31). Certain properties of the organism constitute a parameter that can change its evolutionary trajectory compared to that which its evolution would follow if the organism evolved in a complete causal independence. Even if an organism modifies its environment only slightly, ecological dynamics and selective environments are changed. For example, alleles, which would be unfavourable within a frame of reference without organismic variables, can be fixed in the population (R. Day et al, 20003:88). The more the organism changes its ecological environment, the more it will modify its own selective environment, and the lesser its evolution will track given environmental factors. An example of such an evolutionary situation can be illustrated by the example of the earthworms, whose evolutionary ancestors were aquatic creatures, originally exposed to completely different survival conditions. The quantity of resources available in the environment of earthworms depends on the frequency of the alleles responsible for the presence of digging behaviour, determining the thickness of the arable layer and the quantity of nutritive substances. Therefore, earthworms provide the very resources they exploit. The more alleles responsible for the digging behaviour, the greater the selective pressure on alleles responsible for other properties linked to digging (such as the form of epidermis), the more intense the mucous secretion is. The properties of the organisms can counterbalance and thus neutralize or even change the causal impact of environmental selective pressure. One cannot consider anymore that PARs evolve as a function of environmental factors, and that they therefore reflect this element from the outside world. Another fact is of great interest for the EEB. Organisms generate selection-mediated feedback loop to further evolutionary change of their own lineage (Odling-Smee, et al, 1996, 2003)32. Thus, organisms can “guide” the evolution by generating or re-directing the selective pressure, even when it is operating in the next generations. By transforming its ecological environment at time t1, the organism modifies the selective pressure S, causally affecting the form of its own population features at time t2. For example, spiders’ behaviour on their web constitutes an adaptation to a repetitive factor present in the phylogenetic past, i.e. webs built by those spiders’ ancestors. Once this behavioural variant (web construction) appeared, it generated selective pressure favouring further phylogenetic development of complex morphological and behavioural features constituting a hunting tool. Thus, any anatomical or behavioural variants developing this hunting tool will be retained at t2, t3… reinforcing the selective pressure S generated at time t1 (compare with the evolution of the human brain). Within classical EEB there is one single outside world to which organisms are adapted. The degree of this adaptation varies from one species to another. PARs can give different (as different as species, and more complex animals will have PARs more refined) though never contradictory images of the world. Because living organisms adapt to one and the same world, comparative study of homoplasies or evolutionary convergence between taxa, allows to evaluate the degree of adequacy with the Dinge an Sich (K. Lorenz, 1981:107). Consider now the force of gravitation, which, intuitively, does not seem to be a phenomenon constructed by the organism, but rather a universal physical constraint to which all organisms are subjected, although some more than others. Relatively big vertebrates are more constrained by gravitation than small bacteria floating in liquids, which are rather constrained by the Brownian movement, and bombarded by incessant movement of particles in a thermal agitation. 32

Cf. theoretical models in chapters 3 and 6 of J. Odling-Smee et al., 2003.

32

The organism’s own properties, for example, its size in the case of gravitation, will determine what factor and to what extent will have selective influence on its survival and reproduction. The same is relevant to any other organismic property. Consider the weight of certain Apes. The heavier the animal, the more likely it is to fall down, hurt or even kill itself. Weight exceeding 10 kg represents a serious danger: for Orang-utans, which weigh 40 kg the fall can be easily fatal (D. Povinelli, 1995). For small and light Macaca fascicularis, any support is relatively certain and stable whereas for heavier Orang-utan, the resistance of any support is reduced. Thus, neither organismic, nor environmental properties have evolutionarily absolute value. On the one hand, the fitness of two individuals will vary according to the environmental factors they will be exposed to (one and the same organismic property will not have the same fitness value relatively to two different environmental factors). On the other hand, for one and the same environmental factor, the fitness will depend on the properties of the organism. Selection operates on the fitness (W. Wimsatt, 1981), not on organisms; selective factor is a relative value. Fitness value = f(E,O) Equation is a formal representation of the fitness value, resultant of environmental and organismic variables

Of course, the notion of selective environment does not designate any more a spatially and temporally fixed area that an organism inhabits. Because organisms adapt to the selective environment and the latter is instantiated by environmental AND organismc properties, species are not subjected to the same selective environments and do not evolve according to the same parameters. There is no Dinge an Sich, that could be identified with environmental factors responsible for selective, adaptive origin of RAPs. The Theory of the Red Queen (L. Van Valen, 1973) was the first step in understanding that species are not adapted to an absolute, environmentally fixed value, and that the external selective environment evolves with reference to another. The Theory of Red Queen asserts a certain relativity of the notion of fitness setting out that all increase in fitness of individuals of a (e.g. prey/parasite) species A (good fitness reached by acquisition of more efficient way to dupe individuals of predator/host species B), decreases the fitness of individuals of the concurrent species B. Fitness value is not an absolute since • the selective variable in the frame of reference is not fixed, but changes according to the position of the subject relative to selection entity; • alternately, species A drives selective movement toward species B (species A is a selective cause) and conversely (species B differentiates fitness of species B by acquiring a way to trick species A) Nevertheless, these two variables (species A with reference to species B) determine the fitness value, which alternatively are separated in time. In RTE, these two variables determine the fitness simultaneously.

Conclusion According to the Relative Theory of Evolution, natural selection is not instantiated simply by an external factor. What constitutes selective factor is a resultant of both, environmental and organismic variables. In MS the selective environment represents externally, environmentally determined values, while in RTE, the organism and the environment define the referential of selective environment within which the organism and the environment evolve. The only constant valid in all systems of reference is the viability criterion (note that viability does not imply the externalist view of adaptation, as defined by MS). In RTE, there are two variables in the frame of reference of the selective environment 33

and the change in the value of one (organism) drives the change in the value of the other (environment), and inversely. Selection in this context designates simultaneous and reciprocal causality. In MS, the organism evolves according to some environmentally determined fitness, while in RTE, the organism itself modifies the fitness and the selective factor that it is supposed to adapt to. The organisms do not phylogenetically track an external factor, contrary to MS where natural selection is an asymmetrical process of one way (passive) adaptation of organism to an environmental, independent value. In RTE, organisms evolve without direct reference to some environmental factor - Ding an Sich is not an object but a value, a resultant of environmental and organismic properties. Within MS if there is no known or easy to find correlation between a trait considered as an adaptation and an environmental factor that would be responsible for its selective origin, such a situation is known as evolutionary inertia, and a null hypothesis is considered of explanatory relevance (J. Beatty, 1987), which asserts causal contribution of many factors whose effects compensate. Within the context of RTE the organism is one of the selective causes. The difficulty to establish a direct correlation of environmental factor/adaptation is a natural consequence of the fact that selective factors are never of only environmental origin. What are the consequences for EEB? The Standard Theory of Evolution explains why the organism is adapted to the environment- its form is moulded by natural selection, identified with the selective environment, i.e. factors of the external world having a selective causal importance. Consequently, the Evolutionary Theory of Knowledge explains the existence of a correspondence between the world and PARs- adaptive answers to selective pressures generated by environmental factors. The order of ontological anteriority and causal externalist asymmetry is respected: the environment has the ontological primacy (existing before and independently of the organism) and the causal primacy (constituting the selective cause). From the point of view of adaptationnist externalism, natural selection constitutes the way by which evolutionary modifications occur: a directional change tracks an environmental selective factor. Nevertheless this position neglects the fact that the organism constitutes one of the causal parameters. Fitness (adaptive value) constitutes a dynamic balance, a value resulting from the interaction of these two parameters, external environmental factors and organisms. Adaptation is the function of these two variables, which are both at the same time causes and effects. Within the Relativist Theory of Evolution, adaptation is considered as a process of reciprocal causality (where an organismic change generates an environmental one, and conversely). Neither environment nor organism has either absolute ontological priority (neither of them exists before and independently of the other), or causal priority (neither of them exclusively constitute the selective cause). “Niches do not preexist to the organisms but are constituted according to the nature of the organisms themselves” (R. Lewontin, 2003:8182). One cannot consider the environment as an instantiation of selective environment given that the latter is determined, to some extent, by organism too, and this from the beginning of life on Earth! The organism influences the direction of its own evolution. “Plato’s metaphor of the cave suits here very well. Whatever the autonomous phenomena at work in the external world, they cannot be perceived by [they cannot selectively affect] the organism. Its life [evolutionary fate] is determined by shades on the walls, shades produced by a filter manufactured by himself” (R. Lewontin, 2003:78)33. Within the Standard Theory of Evolution selective pressure is instantiated by the environment. Adaptation derives from Greek ad aptos, subsequently found in latin aptus, 33

La métaphore de la caverne de Platon est ici tout à fait appropriée. Quels que soient les phénomènes autonomes à l’œuvre dans le monde extérieur, ils ne peuvent être perçus par l’organisme. Sa vie est déterminée par les ombres sur les parois, ombres produites par un filtre qu’il a lui-même fabriqué »

34

which means adjusted, appropriate, adapted, made for, and which is in its turn equivalent to fit for in Darwinian writings. Etymologically adapted means: • what is made for something (cf. chapter I); • what is adequate with something. EEB’s quest is similar to St Thomas’ inquiry for adequatio rei et intellectus- truth understood as adequateness between mind’s representation and the world’s state this representation refers to. In EEB, the degree of this adequateness is evaluated in the same manner as in the case of adaptation, by the fitness value measuring the number of offspring reproducing in the next generation. In MS, the environment explains by source laws (E. Sober, 1984) why certain traits are more adapted (have different fitness value) than others in a given environment. Environmental form explains organism’s form. Within the Relativist Theory of Evolution the organism, on the one hand, is the cause of the environmental changes and, on the other hand, differentiates fitness and thus generates selective movement. Consequently, any change in morphology, in behaviour, etc, of the organism constitutes an adaptation to the selective pressure generated and co-determined by itself. In RTE, and Relativist Evolutionary Theory of Knowledge, the number of offspring reproducing in the next generation still constitutes a measure for fitness value. However, the latter does not constitute a measure for the adequateness of PARs to the outside world: the traits considered as “ adaptive responses” correspond to the selective factor, relativist value resultant at the same time from environmental and organismic variables.

General conclusions Does innateness and adaptive origin prove truthfulness of RAPs? If a representation were to relate to a snake only if the latter were present at the moment of ontogenetic (trial/error) learning, the meaning would be available after a contact with the snake, and maybe too late. Thus, the evolutionary raison d’être of PARs is their (partial) innateness, though to be available, PARs must be triggered by ontogenetic experience. Given that natural selection instantiates causal relation, being an adaptation guarantees the relation of correspondence between selective factors and PARs (feature/adaptations, equivalent with the Kantian notion of synthetic a priori knowledge) conferring on the latter the status of knowledge. In our analysis, did we prove that A knows a priori that p because the belief of A that p is justified not by learning, but by phylogenetic experience? If A have an innate belief that snakes are dangerous, this is because in the majority of contacts snakes/A’s’ ancestors snakes were venomous. However, even if we suppose continuous control of stabilizing natural selection toward PARs, it does not change the fact, that if A has an innate belief that snakes are dangerous, its belief will be true not be the virtue of its innateness, but will be true only if the snake it scarcely perceives and of which it is distrustful, is actually venomous. Does genetics of population have means to establish causal (selective and phylogenetic) relations? PARs were considered as modular units, or quasi-independent associations of genetic and phenotypic properties. PARs are associations gene/trait, conditions for adaptive evolution (R. Dawkins, 1982; J. Bonner, 1988). In this sense, PARs are modules traits/genes matches weakly connected to other matches traits/genes with mainly inner pleiotropy (R. Brandon, 1999). Those relatively fixed associations result from pleiotropic and epistatic relations, commonly inherited thanks to genetic linkage and recessivity. PARs or modules are units of inheritance hardly divisible by natural selection because they have a 35

common ‘function’, that constitutes a ‘response’ to a common selective pressure. There is a supposed correspondence between common selective pressure (set of environmental factors) exerting its action on common function (set of organismic properties). It implies that modules keep their quasi independence given that natural selection acts on the traits without altering other traits of the organism in a countervailing fashion (R. Lewontin, 1978:230). Thus PARs correspond to • homologies or symplesiomorphic properties (features shared by related species) identified in systematic cladistic studies • specialised functions, identified through reverse engineering analysis

Selective pressure → [ functional unit →

(phenotypic traits → genes)

]

Natural selection is a change of gene frequency through phenotypic traits (having a function corresponding to en environmental pressure), and through them affects the genetic basis of the latter (R. Brandon, 1999).

In the first chapter we show that the notions of function and adaptation do not overlap. The first results from an ahistorical, engineering-like analysis, where evolution by natural selection is considered rather as the process of artificial selection, where organisms are considered as artefacts, and where from benefit or functionality and usefulness of some properties, adaptations are deduced. We show the relative circularity of this practice and examples where through rational engineer’s thought evolutionary cause cannot be identified, which denies the possibility of rational (reverse engineering) deduction of selective events responsible for origin of adaptations. In this perspective, the notion of proper direct function seems misleading. In the second chapter we show a set of a priori (intuitive or folk) concepts of biology and the conceptual evolution of the latter leading to the inconsistencies enabling us to go beyond them; we try to show that modern biology is undergoing a revolution, as geometry and physics have had in the last hundred years. We first analyse such inconsistencies in taxonomy. The first one appears thanks to the phyletic species concept, combined with the anagenetic model of speciation, where the emergence of new species (of new quality) does not imply new numerical identity. We show that the biological concept of species and cladogenetic model of speciation are based on the notion of substance (or essence), corresponding to universals for humans folk essentialism and folk biology. Substance (species) constitutes a PAR; thus, all arguments proving its objectivity by its adaptive origin (e.g. J. Maynard-Smith) are circular. The second inconsistency appears in systematics with the exclusivity (asymmetric or not) between explanation of common origin and of adaptive origin, that is in the identification of homoplasies and synapomorphic properties (not inherited from common ancestor). Quest for evolutionary convergences (considered as adaptations) shows another a priori concept, that of common cause explanation (deduction from two correlated properties of the contribution of one common cause, rather than contribution of two independent causes). The quest for continuous control of stabilizing natural selection and adaptations in proprieties that are clearly homologies and symplesiomorphic properties (properties shared by related species) shows a priori character of causal explanation tout court. Then we assert that thanks to the genetics of population we can see that natural selection, or drift, have not necessarily their equivalent in a mind-independent entity or group of entities. Consequently, it is no possible to establish the cause of evolutionary events. Many possible explanations are consistent and provide a good predictability, but none is contextindependent. We temper our realist pretensions and assent to the following criterion of causal 36

identification: “if the fitness of x is context dependent, then either (1) selection is acting upon a level of organisation higher than x; or (2) there are several different selection processes occurring at the level of x” (K. Waters, 1991:558). This shows another mark of conceptual revolution in biology. One of Euclid’s axioms asserts that if we consider a point and a straight-line, there exists only one straight-line parallel to this straight-line, and crossing the point. Non-Euclidian geometries deny this axiom asserting that either there is no straight-line parallel to this straight-line, or there exist many of them. It does not mean that Euclidian geometry is true, but that our a priori geometrical knowledge is not the only one possible and ensuring a viability condition. There is not possibility to prove through PARs that ordo et connexio idearum idem est, ac ordo et connexio rerum. It is still far from certain that the folk metaphysic notions underpinning the Modern Synthesis actually reflect a mind-independent phenomena. “For if we regard outer appearances as representations produced in us by their objects, and if these objects be things existing in themselves outside us, it is indeed impossible to see how we can come to know the existence of the objects otherwise than by inference from the effect to the cause; and this being so, it must always remain doubtful whether the cause in question be in us or outside us (I. Kant, 1980:A372). What is certain, however, is that there is another consistent way of conceiving reality, at least as efficient as that dictated by folk concepts. Circularity in proving veracity of PARs would not be vicious (G. Vollmer, 2005), if we could defend what follows: a priori concepts are adaptations; the degree of this adaptation varies from one species to another; PARs give different (as different as species are) but never contradictory representations of the world. Nevertheless, as the analysis of chapter 3 shows, the only ‘objective’ criterion of veracity is that of viability, and that criterion does not imply the externalist view of adaptation, according to which living organisms adapt to one and the same world, world meaning selective environment! Indeed, in chapter 3 we deconstruct the evolutionary externalism of Modern Synthesis and conclude, by introducing the Relativist Theory of Evolution, that selection operates on the fitness and not on organisms, that the latter is a resultant of environmental and organismic variables, and finally, that selective factor is a relative value. Thus, RAPs do not refer to absolute external values or physical environmental objects (Dinge an Sich from “the ultimate external reality”, M. Ruse, 1990), but to the relational values, resultant from the relation between environmental and organismic properties. What does relativist in Relativist Theory of Evolution mean? Einstein’s theory of relativity does not states that things are relative to the observer, but to the point of reference (A. Murphy, 1927). The same goes for RTE: the ‘objects’ to which PARs refer are, however, relative and objective properties. “In a simpler world, the process that produces adaptation would be certain to increase the level of adaptedness. In a simpler world, an organism’s “adapting” to its environment by modifying its behaviour and physiology would be the same kind of process as population’s “adapting” to its environment by modifying its genetic composition. If matters were simple, frequency-dependent selection would not occur and evolutionary adaptation would have the same structure as ontogenetic adaptation” (E. Sober, 1984:211).

37