by Greg Ransom

This paper was written for Alexander Rosenberg and Larry Wright
in partial fulfillment of the Ph.D requirements at UC-Riverside.


“When we descend to details we can prove that no one species has changed; nor can we prove that the supposed changes are beneficial, which is the groundwork of the theory.”
--  Charles Darwin


One of the central claims of my paper is that within the complex sciences (like evolutionary biology) we must be satisfied with the prediction and explanation of the abstract characteristics of a general pattern of events -- and we can never hope for anything more precise than a multiply-realized and often open-ended or incomplete how-possible causal explanation of a particular event appropriately identified as falling within a complex science (compare Hayek, 1967; Prigogine, 1985; Kincaid, 1988; Stone, 1989; Kincaid, 1990; Brandon, 1990).  My conclusion is that the model of science which looks to the subsumption of particulars under causal laws as the mark of scientific truth and metaphysical reality is inadequate for those sciences which rely upon the direct perception of patterns which defy adequate or exhaustive description in purely causal terms.

There is, however, a presumptive philosophical position which asserts that teleological explanations and ascriptions in the biological sciences are scientifically legitimate only because as a matter of principle these explanations may be reduced to a completed set of causal explanations by scientists who possess sufficiently expansive cognitive resources (compare Rosenberg, 1982; Kincaid, 1988; Rosenberg, 1989).  It is my aim in this paper to establish that no such reductions are in principle possible in the biological sciences due at once to innate characteristics of our human perceptual faculties and due concurrently to certain features which are intrinsic to the structure of biological explanation. 

In spite of these insuperable limitations to causal reduction in the biological sciences, I wish defend the legitimacy of teleological explanation in biology against those who would give privileged status to causal explanation in the physical sciences.  Toward this end, I intend to identify and account for the explanatory power of teleological ascriptions by characterizing scientific explanation itself as a process of pattern modeling.  I plan, what is more, to argue for the independent and secure status of teleological explanations in biology by pointing to the well-established reliability and overwhelming intersubjectivity of our teleological perceptions which I claim suffers nothing in comparison to the reliability and intersubjectivity of our perceptions of constant conjunction.

In this paper I also propose an original characterization of the structure of teleological phenomena and suggest a somewhat novel account of our relation as human beings to such phenomena.  My account of teleological phenomena will unfold through the course of the middle section of this paper.  There are, of course, a number of important desiderata by which to judge the philosophical adequacy of any proposed construal of our linguistic and scientific practices with regard to teleological phenomena.  For my own money, I believe that an adequate construal of teleological explanation and ascription should be able to both characterize and account for each of the following:  1) our ability to directly comprehend and perceive teleological phenomena without need for higher intellectual reflection; 2) erroneous teleological ascriptions; 3) the analogy between human purposes and artifacts and biological adaptations; and 4) metaphysical worries about the subjective or chimerical status of teleological ascriptions. 

In the following pages I attempt to show how my account of teleology can successfully address these difficult concerns without conceding to the in principle reducibility or normatively mandated eliminatibility of teleological ascriptions and explanations to the 'causal laws' of physics or chemistry.


There is, of course, an empiricist tradition which holds that the sciences seek to explain particular events by bringing those events under the scope of universal laws.  (Compare Scheffler, 1963, p. vii; Popper, 1968, p. 59; Salmon, 1990, p. 8;  for difficulties with this model of scientific explanation, see Hayek, 1952; Suppe, 1988; Giere, 1988; Van Fraassen, 1989; and Salmon, 1990).  A long-standing challenge for empiricists in this tradition has been to reconcile their conception of scientific explanation with the ubiquitous use of teleological attributions and explanations in the biological and human sciences.  A common tack taken by many empiricists is to demand the full rejection or elimination of teleological attributions and explanations unless it can be show how these attributions and explanations might be successfully translated or reduced to non-teleological laws and descriptions.

On this view of scientific explanation, without the open possibility of translation or reduction to causally defined initial conditions and sets of lower-level universal laws, all teleological attributions or explanations are to be rejected as either metaphysically suspect or scientifically illegitimate.  Perhaps not unsurprisingly, empiricists in this tradition often identify scientific progress with what they perceive as the linear historical trend toward the elimination of teleological explanations from science in favor of causal laws.

The explanatory scruples of this empiricist tradition are driven by three closely related concerns.  These concerns are with:  (1) the unity of science; (2) the objectivity of science; and (3) the metaphysical economy of science.  These concerns are closely related in that all three concerns seek in one fashion or another to capture our sense of science as giving us an accurate model of the world as it exists independently of ourselves.  They are also related in that each  -- in its own unique way -- reflects our everyday experience that almost never does it happen that something comes from nothing.

It is important to recognize that these concerns are not exclusive to any particular philosophical tradition -- empiricist or otherwise.  In fact, all three concerns help independently to ground our modern rejection of the animistic explanations of physical phenomena which were once provided by primitive religion.  A tribal priest might explain the movement of a rock as an expression of the purposive desires of the conscious agent who is the rock.  Today, with more powerful and informative physical explanations at hand, we typically wish to explain away these animistic explanations as an illegitimate projection of our own anthropomorphic characteristics into inanimate systems or bodies. 

In hindsight, the scientific inadequacy of this projection is revealed by its failure to meet the implicit tests for scientific legitimacy found in the three concerns I provided just above.  First, it is revealed in the fact that animistic attributions are relative to particular individuals and societies -- they are not intersubjectively shared across persons and cultures.  As a result, animistic explanations fail the test of scientific objectivity.  Second, it is revealed in the fact that animistic explanations float free from the constraints of the underlying elements and mechanisms which constitute the phenomena under explanation -- they are not bound by any of the gross properties of their constituents.  As a result, animistic explanations fail the test of the unity of science.  Finally, it is revealed in the fact that animistic explanations require the reification of conscious entities for which for which there is little or no evidence and  -- especially in light of other and more powerful explanations -- for which there is no need.  In effect, under such conditions animistic explanations are simply superfluous window dressing which are unable to carry important explanatory weight.  As a result, animistic explanations fail the test of metaphysical economy.   

Of course, it should be recognized that animistic explanations are a form of teleological explanation.  Indeed, for a great many empiricists it is the dismaying example of animism itself which casts a dark pall over all forms of teleological explanation in every branch of scientific inquiry.  The argument here -- if we deem call it such -- is one from analogy.  The concern is that all teleological ascriptions may project on to objects and processes a property or properties which objects and processes do not themselves possess, but which in fact are possessed only by the observer -- if they are in fact possessed by anything at all (compare Woodfield, 1976, p. 26).  The suggestion, then, is that all teleological explanations may be anthropomorphic explanations -- and that all such explanations may be unwarranted projections as dubious as those which are offered by a tribal priest. 

In fact, many empiricists infer the dubiety of all teleological explanations by a process of inductive extrapolation from what they view as the progressive elimination of animistic and other forms of anthropomorphic explanation from the study of nature.  On this view, the progress of science itself both demands and implies the elimination of teleological explanations from science.  Unfortunately, however, empiricists who insist upon proscribing the use of teleological and other anthropomorphic ascriptions to characterize and explain the behavior of not only themselves but also other men would seem to be at a complete loss to account for the very phenomena they condemn -- i.e. the mistaken attribution of anthropomorphic characteristics to inanimate phenomena (compare Hayek, 1948, p. 65). 


Many empiricists have attempted to capture our sense of the objectivity of science by locating that objectivity in the intersubjectively shared observation of bare particulars and their constant conjunctions.  On this view -- which begins with a controversial presumption that science seeks to explain particular events by subsuming those events under universal laws -- only the identification of causal regularities can account for what we commonly recognize as the objectivity of scientific explanation.  Indeed,  according to this account, it is just because the objectivity of science is located in our intersubjective experience of universally conjoined particulars which allows science to be put to a test which is not relative to an individual or a cultural group.  (For some of the difficulties with this view of scientific objectivity, please see Polanyi, 1958; Hanson, 1958; Kuhn, 1970; Kuhn, 1977; and Newton-Smith, 1981).

This characterization of the objectivity of science can be found in the work of both Carl Hempel and Karl Popper, among many other philosophers.  According to Hempel's account, "..scientific research seeks to account for empirical phenomena by means of laws and theories which are objective in the sense that their empirical implications and their evidential support are independent of what particular individuals happen to test or to apply them; and the explanations, as well as the predictions, based on such laws and theories are meant to be objective in an analogous sense."  (Hempel,      ,  p.     )  According to Popper's account, "..the objectivity of scientific statements lies in the fact that they can be inter-subjectively tested .. Kant was perhaps the first to realize that the objectivity of scientific statements is closely connected with the construction of theories -- with the use of hypotheses and universal statements.  Only when certain events recur in accordance with rules or regularities, as is the case with repeatable experiments, can our observations be tested -- in principle -- by anyone .. Only by such repetitions can we convince ourselves that we are not dealing with a merely isolated 'coincidence', but with events which, on account of their regularity and reproducibility, are in principle inter-subjectively testable."  (Popper, 1969, p. 44-45)  

Perhaps unsurprisingly, a particular conception of the demands of objectivity in science has led many empiricists to a particular conception of the purposes of sciences.  According to this view, the purpose of science is to provide us with the linguistic apparatus necessary to derived intersubjectively shared perceptual implications from our past experience of the world.  As Hempel puts it, "science .. is concerned to develop a conception of the world that has a clear, logical bearing on our experience and is thus capable of objective test."  (Hempel, 1966, pp. 47-48)

How sound is this view?  I would suggest that the initial plausibility of this account of scientific objectivity must be tempered by consideration of several noteworthy imperfections.  It would appear that science on this view may always be characterized as a tool for predicting particular experiences which may be deduced from past experience -- without any need for speaking of an external world.  This result may be seen as a symptom of the empiricist's general inability to provide us with a coherent picture of what it means to have an experience of the world.  It must be noted, therefore, that the presentationalist picture of science which derives from the constant conjunction model of scientific objectivity stands in direct tension with precisely those intuitions about science (e.g. that science is a shared model of the world as it exists independently of ourselves) which in fact provokes our search for the formal grounds which stand behind and lend legitimacy to what we routinely perceive as the objectivity of science. (Compare Hayek, 1952; Quine, 1953; Carnap, 1956; Rorty, 1979; Popper, 1983; Putnam, 1983; Bartley, 1987; Baker, 1987; Edelman, 1989.)

In fact, there is good reason to insist that the constant conjunction interpretation of science is incompatible with the truth of many of the most secure hypotheses of science.  It has been argued, for example, that such an interpretation of science cannot accommodate the theory of natural selection because it cannot be squared with the truth of the assertion that scientists (i.e. the possessors of scientific truth) are themselves a product of the process of natural selection.  (Compare Hayek, 1952; Hayek; 1967; Lorenz, 1977; Clark, 1983; Millikan, 1984; Clark, 1986; Ruse, 1986; Bradie, 1986, pp. 444-451; Bartley, 1987; and Edelman, 1989.)

What empiricists need to show us is how their account of scientific objectivity can retain the commonsense notion that our sensory experiences are of a world which exists independently of ourselves while at the same time accommodating the fact that this world may be known both crudely through the models provided by sense experience as well as more precisely through the models provided by science.  One of the great difficulties here is for the empiricist to make sense of the fact that our conception of the world both begins with and is premised upon the distinction we are each intuitively able to make between the influence which the world has upon ourselves and the influence which each of us have in turn upon the world.  Let me suggest that this distinction has been brought into contemporary prominence by scientists and philosophers who have recognized that a sound interpretation of science must accommodate the prominent role which human purposes, expectations, plans, and dispositions play in the process of science -- including here both the acquisition of language and the interpretation of experience.  Perhaps more significantly it is also increasingly recognized that these capacities cannot with coherence be reduced to or explained in terms of the constant conjunction model of science.  (See Hayek, 1952; Hayek, 1967; Popper, 1962; Popper, 1974; Popper, 1983; Putnam, 1983; Baker, 1987; Edelman, 1989.  Compare also Wittgenstein; 1953; Wittgenstein, 1958; Mises, 1976; Quine, 1960; Quine, 1969; Quine, 1990; Kuhn, 1970; Kuhn, 1977.) 

It must be admitted, of course, that the empiricist's claim that only a causal model of scientific explanation can provide us with an objective standard of scientific truth does seem to receive strong intuitive support from the high frequency of intersubjective agreement which individuals report in their collective experience of constantly conjoined particulars.  Yet for many empiricists the most important support for causation as the objective basis of scientific explanation is thought to be found in the very structure of human language.  (On the linguistic basis of modern theories of objectivity see Johnson, 1987, pp. ix-xiii, pp. xxii-xxiv, and p. 148; and Lakoff, 1987, pp. 157-184). 

For example, according David Hume, "Among different languages, even when we cannot suspect the least connection or communication, it is found that the words expressive of ideas the most compounded do yet nearly correspond to each other -- a certain proof that the simple ideas comprehended in the compound ones were bound together by some universal principle [i.e. the principle of constantly associated particulars] which had an equal influence on all mankind."  (Hume, 1955, p. 31-32)  According to this view -- especially as it has been by advocated by contemporary empiricists -- it is the very structure of language which reveals the underlying formal requirement for providing objective scientific explanations.  (Compare, for example, Quine, 1960; Quine, 1969; Quine, 1990).

Given this linguistic background, it is not surprising that many empiricists are ready to impugn the legitimacy of explanatory teleological ascriptions on the grounds that such explanations and ascriptions cannot provide us with objective criteria for testing their appropriate linguistic use.  For example, according to Hempel, a serious flaw with functional explanations in sociology and biology  ".. consists in using key terms of functional analysis, such as 'need' and 'adequate (proper) functioning' in a nonempirical manner, i.e., without giving them a clear 'operational definition,' or more generally, without specifying objective criteria of application for them."  (Hempel, 1965, p. 319-320). 

The perception that teleological attributions and explanations often make use of words which lack starkly defined and intersubjectively testable observational implications -- a perception which is only accentuated by the stark contrast with the empiricists explanatory paradigm of causal explanation -- serves to undermine the scientific status of these explanations for many empiricists.  As Hempel goes on to argue, "If functionalist terms are used in this manner, then the sentences containing them have no clear empirical meaning; they lead to no specific predictions and thus cannot be put to an objective test; nor, or course, can they be used for explanatory purposes .. Nonempirical use of functionalist terms may, therefore, bar sentences of these various kinds from the status of scientific hypotheses."  (Hempel, 1965, p. 320)

Ernest Nagel is another empiricist who is deeply concerned about the objectivity of teleological explanations.  Nagel is specifically interested in the scientific status of the distinction biologists and others perceive between goal-directed and non-goal-directed processes.  As an empiricist, Nagel naturally wishes to provide a purely causal account of goal-directed systems which captures the intuitively perceived distinction between goal-directed and non-goal-directed phenomena within what he conceives of as the 'objective' language of bare particulars and constant conjunction. 

The perceived need to identify objective criteria which may be used to reliably distinguish between goal-directed and non-goal-direct processes is therefore a major motivation behind Nagel's own demand for the translation of teleological language to non-teleological language.  As Nagel later explained, an important aim behind his own system-property view of goal-directed phenomena, "was to produce an objective and reasonably precise characterization of what it is that distinguishes processes in biology that supposedly are goal-directed from those which are not."  (Nagel, 1977, p. 272) 

It is no accident, therefore, that we find that one of the central demands of Nagel's own system-property view of teleological behavior it that the causal variables relevant to the realization of a system's goal state must mutually independent of one another.  According to Nagel, this requirement is of central importance because it serves as a formal criterion for distinguishing processes that are goal-directed from those which are commonly held not to be such.  (Nagel, 1977, p. 274)

Nagel's demand that teleological explanations should be reducible to non-teleological explanations is therefore a response to his concern for the objectivity of science.  A major aim of the reduction is to provide objective warrant for our attributions of goal-directedness.  As an empiricist in the constant conjunction tradition, Nagel's strategy is to provide an objective causal standard from which to adjudicate legitimate from illegitimate attributions of goal-directedness.  In effect, then, the independent variable requirement is intended to provide Nagel with the causal framework he needs to derive what to his lights is an objectively testable criterion for goal-directedness.     

It almost goes without saying, of course, that empiricists are also intensely interested in the metaphysical economy of science.  There is, in fact, an empiricist tradition which objects to the positing of beings, entities or forces which exist independently of any causal nexus.  On this view, the condition for the existence of an event is that it be preceded by a constantly conjoined prior event -- i.e. that it be caused.  As Hume concisely puts the point, "if the first object had not been, the second never had existed."  (Hume, 1955, p. 87)  Once again, such a concern is often closely tied to particular views about the nature of language, as we shall shortly see.

Before I turn to a more explicitly linguistic example, however, let me first quickly dispose of several of the most obvious objections to teleological explanations -- objections which an not exclusive to empiricists of any particular tradition.  According to Andrew Woodfield, "The most common criticisms of teleological explanations nowadays are either that they are animistic, i.e., they assume that the thing being explained has a mind, or that they tacitly invoke a supernatural being who directs the course of events."  (Woodfield, 1976, p. 3)  Yet, as Larry Wright persuasively argues, the explanatory force of anthropomorphic metaphor in no way depends upon the attribution of conscious volition to teleologically characterized phenomena.  According to Wright, a good anthropomorphic metaphor can organize and highlight our perceptions in a fully explanatory and intersubjectively testable fashion without any need to import the uneconomical excess metaphysical baggage represented by the attribution of conscious volition to inanimate phenomena.  (Wright, 1976, pp. 12-22.) 

Although many would not accept all of the details of Wright's account, most contemporary empiricists also reject this form of the anthropomorphic objection to teleological explanation for reasons very like those offered by Wright.  As Ernest Nagel explains, "It is a mistaken supposition that teleological explanations are intelligible only if the things and activities so explained are conscious agents or the products of such of such agents."  (Nagel, 1979, p. 24)

A second common objection to teleological explanation is that this form of explanation is dependent upon a backwards form of causation in which the future act as an agent upon the past.  As Alexander Rosenberg explains, "Traditionally, it is of course the teleological statements' reference to future states which, it purports, somehow determine past ones, that has rendered its claims suspect."  (Rosenberg, 1980, p. 54.) 

The claim here is that teleological explanations are committed to a metaphysically extravagant notion of causation in which future events cause past events.  However, as Andrew Woodfield persuasively argues, "Teleological explanation cannot possibly be committed to any reversed causal hypothesis .. [because] teleological descriptions are capable of being true whether or not the event cited as the purpose occurs."  (Woodfield, 1976, p. 34.)  For this reason it is simply a mistake to conceive of teleological explanation as a form of causal explanation.  Indeed, to do so would require us to produce a wildly uneconomical metaphysics in which non-actual future events often act as the cause of already previously determined events. 

It is unsurprising, therefore, that few empiricists today argue that teleological explanations are committed to the existence of forces which act backwards in time.  As Ernest Nagel explains, "It is .. a mistake to suppose that, because teleological explanations contain reference to the future in accounting for what already exists, such explanation must tacitly assume that the future acts causally on the present."  (Nagel, 1979, p. 24)

Despite the common admission of contemporary empiricists that teleological explanations are not committed either future causation or animism, the reaction of many modern empiricists to the animistic biology of the German biologist Hans Driesch is still quite telling for what it reveals about the lessons which empiricists have learned from familiar examples of metaphysically extravagant teleological explanations.  Driesch, a turn of the century developmental biologist, posited the existence of a non-physical teleological force which caused biological organisms to develop with the future in mind.  Driesch called this force an 'entelechy'.  Rather than simply objecting to the metaphysical extravagance of Driesch's account of developmental biology, the response of many empiricists was to deny the very meaningfulness of Driesch's explanatory claims concerning the role of an entelechy in the regeneration of body parts.  For example, Rudolf Carnap recounts the reaction of his own philosophical group to Driesch's metaphysically extravagant explanation of teleological phenomena.  According to Carnap, "We said to him:  'Your entelechy -- we do not know what you mean by it.'"  (Carnap, 1966, p. 14) 

The metaphysical lesson learned by Carnap and many other empiricists from Driesch's example was a lesson about the nature of language, and from this linguistic lesson Carnap and many other empiricists inferred what the took to be the metaphysically appropriate form in which to express scientific explanations.  The supposed lesson of Driesch's entelechy, as Carnap explained, was that, ".. you cannot give an explanation without also giving a law."  (Carnap, 1966, p. 14)  This unstartling conclusion was, of course, fully consistent with the already commonplace view of many empiricists which insisted that all science seeks to explain particular events by subsuming those events under universal laws.  The practical result of Driesch's lesson for empiricists like Carnap was to mandate the enforcement of metaphysical economy by means of simple test for linguistic form. 


An empiricist's concern for the unity of science is often deeply conditioned both by a constant conjunction model of scientific objectivity and by a formal linguistic test for metaphysical economy.  This can be seen in an empiricist's frequent expression of the concern for the unity of science as a methodological issue dealing with the proper scope and form of scientific theories and their logical attributes and interrelations.  Yet at a deeper level, the empiricist's concern for the unity of science reflects a common intuition about the additive properties of mereological sums.  This is the intuition which suggests that the properties of a brick wall are produced by or are a product of nothing other than the properties of the bricks and mortar which make up the wall.  Although certainly not unchallenged by both scientists and philosophers, this intuition appears quite sound when we reflect upon the fact that a brick wall is made up of nothing but bricks and mortar -- and given our background experience that rarely or never does something come from nothing.

The same sort of intuition supports the commonsense view of the unity of science which holds that biological organisms consist of -- or are nothing but -- chemical systems and their physical components (Schaffner, 1969, p. 346; compare Kim, 1978, p. 155).  It is, however, no small step from this commonsense view of the unity of science to the claim made by a number of empiricists that, " .. [their] commitment to the unity of science obliges [empiricists] either to show how teleological language can be analyzed into the non-teleological language of physical science, or to read teleological biology out of the natural sciences altogether."  (Rosenberg, 1982, p. 161; compare Beauchamp and Rosenberg, 1981, pp. 308-309, pp. 317-321).

Typically, of course, this reductive test for the scientific status of teleological ascriptions and explanations is tempered by an 'in principle' caveat which provides special dispensation for those teleological explanations and ascriptions which have yet to be reduced to chemistry and physics.  This in principle caveat allows that while a completed physical reduction in any particular instance may be beyond the ken of man due to the computational limitations of the human brain, it still remains true that a specific physical reduction must -- in principle -- be available to a hypothetical individual possessing the necessary intellectual resources (see Kincaid, 1987; Kincaid, 1988; and compare Rosenberg, 1984; Rosenberg, 1985; Rosenberg, 1989).   

We can fairly say, then, that the empiricist's reductive test for scientific legitimacy insists that for any particular teleological ascription or explanation it must be possible -- at least in principle -- to render that explanation or ascription in terms of some lower-level disjunction of universal laws and physical characterized items -- and this reductive standard (as Harold Kincaid suggests), "..minimally requires that someone or some group of people with sufficient computational ability could provide the relevant disjunct."  (Kincaid, 1988, p. 273).  But it is just this claim, I shall argue below, which cannot be sustained.

First, however, I would like to suggest that a denial of the claim that a physical reduction of teleological ascriptions and explanation may always be provided in principle by some hypothetical superbrain is no denial of the intuitions which lie behind our concerns for the unity of science.  While it may well be true (as I hope to demonstrate) that no scientist -- however cognitively well-endowed -- could have access to a closed and yet complete physical disjunct for every scientifically legitimate teleological explanation, it need not follow that, for every particular bit of teleological phenomena previously observed by a man, there is not some particular constellation of law abiding or physically characterized items which constitute that bit of phenomena. 

My claim, then, is that while we may wish to hold -- for whatever reason -- that a bit of teleological phenomena is made up of physical parts and nothing more, it fails to follow that we can derive the physical constitution of all those disjuncts which might fall into a particular teleological category.  In fact, we might correctly assume that a particular bit of teleological phenomena has a particular physical constitution without ever being in a position to specify a finite set of possible disjuncts which includes that particular physical configuration.  Indeed, it may well be the case that we cannot possess the criteria required to circumscribe the bounds of such a set.  A difficulty of this sort would effectively constitute an in principle limitation to physical reduction in spite of the previously assumed truth of mereological determinism. 

It follows, therefore, that the truth of mereological determinism does not establish or entail reducibility in the sense previously specified (compare Kim, 1978, p. 154; Kincaid, 1987; Kincaid, 1988).  It also follows, conversely, that the existence of an insuperable barrier to reduction does not establish or entail the falsity of mereological determinism.  As Kincaid explains, "While we may never be able to define biological predicates, for example, in chemical terms, we can still hold with good reason that the chemical facts fix the biological facts and that every biological event is brought about by or realized in chemical phenomena."  (Kincaid, 1988, p. 251)        

The existence of such a possibility shows that a demand for the in principle reducibility of teleological ascriptions and explanations to physics and chemistry is a demand which goes beyond what is implied by a metaphysical commitment to mereological determinism.  Mereological determinism is a metaphysical claim about the relationship between parts and wholes whose truth implies the supervenience of upper-level phenomena upon lower-level phenomena.  Its truth, as we have seen, does not imply that, "..once lower-level theories are fully developed they would suffice to derive all adequate higher-level explanations -- if we removed all limitations of time and resources."  (Kincaid, 1990, p. 576) 

In fact, such an assertion -- which is implied by the demand for the in principle possibility of teleological reduction -- must be considered a hypothesis about the sufficiency of particular lower-level explanations to account for some higher-level phenomena.  This later hypothesis may in fact be quite false despite the truth of mereological determinism.  We might, for example, have good reasons to doubt the scientific claim while still holding to the truth of the metaphysical claim.  As Kincaid suggests, "Knowing that supervenience is true is far different from possessing a function to generate all possible supervenience bases."  (Kincaid, 1988, p. 273; compare Kincaid, 1987; and Kim, 1978)       


Before I attempt to establish that there do in fact exist insuperable limits to teleological reduction in biology, it is important first to take a closer look at the nature of the phenomena which is the subject of our discussion.  Perhaps the most striking feature which is characteristic of teleological phenomena is that such phenomena may be directly perceived by men without resort to higher intellectual operations.  In fact, our ability to intuitively recognize teleological patterns is one of our most basic human skills.  Empiricists themselves can be found speaking of "manifestly goal-directed behavior" and of "observable teleology" (see Nagel, 1979, p. 401-402; and Rosenberg, 1985, p. 65).  

It is worth pointing out that we are frequently able to perceive purposes directly although we may be unable to specify precisely how we have done so.  As Friedrich Hayek correctly remarks, "we recognize these patterns from we do not know what." (Hayek, 1967, p. 53)  It is no accident, then, that we are usually able to recognize teleological patterns intuitively long before we can describe these patterns in specific terms (compare Hayek, 1967, p. 55).  In fact, the perception of teleological phenomena almost always precedes its precise characterization.       

The direct and intuitive comprehensibility of teleological phenomena is routinely confirmed in both classical and contemporary biological literature.  According to Charles Darwin, ".. the same reasoning power which tells us plainly that most parts and organs are exquisitely adapted for certain purposes, tells us with equal plainness that . . rudimentary or atrophied organs, are imperfect and useless."  (Darwin, 1964, p. 453)  According to George Williams, "It is often easy, in practice, to perceive functional design intuitively."  (Williams, 1966, p. 9)  Later Williams again repeats the claim that, "functional design . . can be intuitively comprehended."  (Williams, 1966, p. 260)  Intuitive attributions of design are a common place throughout the field of evolutionary biology.  As Charles Darwin himself explains, "Nothing can be plainer than that wings are formed for flight ..".  (Darwin, 1964, p. 451)  Let me suggest that if the scientific authority of our most experienced and respected biologists is to count for anything we must assume that these plainly observed adaptations and functional designs are to be included among "the manifest properties of organisms" (Rosenberg, 1984, p. 108). 

In fact, teleological patterns are such a part of our lives that we take them for granted.  As a result, the evident teleology of plants and animals causes little surprise and raises few questions for acting man.  What happens is that we observes patterns which resemble the familiar patterns of our own conscious behavior.  These patterns seem as little in need of explanation as do our own actions.  This situation changes radically, however, after we have identified the origin of species as a problem to be explained.  In other words, only after we have admitted to ourselves that the plants and animals of the world have been created or produced by some other agent or process we are then left to wonder how each individual plant or animal has been endowed to display teleological patterns.

The origin of species by modification and descent seems like a quite plausible answer to the problem raised by the emergence of species.  This account of the origin of species is hopeless, however, if it serves merely to generate far greater difficulties than it attempts to solve.  The primary problem fathered by theories of species origin is the problem presented by the manifest teleology of biological organisms and their many parts.  In order to be scientifically viable, a theory of species origin by modification and descent must explain how plants, animals and other organisms have been endowed by nature with manifestly teleological properties.  As Charles Darwin explains, " .. such a conclusion [the origin of species by descent], even if well founded, would be unsatisfactory, until it could be shown how the innumerable species inhabiting this world would have been modified, so as to acquire that perfection of structure and coadaptation which most justly excites our admiration."  (Darwin, 1964, p. 3) 

Let me suggest that the situation in evolutionary biology is very like the situation described by Larry Wright in his characterization of teleology as a philosophical problem.  In biology, as well as philosophy, there are those who wish to take an eliminativist position with regard to teleology.  It is argued by some that physical theory can account for the mere 'appearance' of teleology and that such appearances may be set aside as mistaken or illusory.  For example, what most of us acknowledge to be the goal-directed behavior of men and animals is described as only 'apparently' goal-directed by these theorists.  Wright's candid response to this approach is simply to highlight the direct observability of teleological behavior.  In this way Wright forces his opponent to acknowledge the phenomena which first provokes our interest and still asks to be understood.  As Wright explains, "In a large number of cases, the goal-directedness of a bit of behavior is obvious on its face .. Occasionally there is simply no question about it:  the rabbit is fleeing, the cat stalking, the squirrel building a nest .. ".  Wright adds, "..  it is because of this that we have any reason at all to think that there is anything to give a philosophical analysis of."  (Wright, 1972, p. 206) 

I wish to claim that the situation in evolutionary biology is very much the same.  It is the manifest teleology of biological organisms which in the first instance has presented evolutionary theorists with the phenomena they desire to explain.  Only later, and after great success with the explanation of teleological phenomena, did biologist turn to the explanation of very complex and yet ubiquitous structural features such as the lateral line in fishes.  The starting point for evolutionary biology, however, is the identification of, "some feature of the organism -- morphological, physiological, or behavioral . . [which] serves some proximate end (food getting, escape, etc.) that the observer believes he can discern fully by direct observation and without reference to the history of the organism."  (Colin Pettendrigh quoted in Williams, 1966, p. 258) 

In fact, it is precisely this sense of adaptation which first provokes the biologist.  It is no accident, then, that biologists since the time of Darwin have recognized the primary importance of teleological phenomena for evolutionary theory.  As John Maynard Smith explains, ".. adaptation is the most obvious and all-pervasive feature of living things, and one that any theory of evolution must explain."  (Maynard Smith, 1989, p. 4) 

Significantly, the centrality of direct teleological perception and intuitive teleological comprehension to evolutionary biology places philosophical concern over the objectivity of teleological explanation into the heart of the biological sciences.  This concern is not unknown to contemporary biologists.  Richard Lewontin, for example, points out that a serious difficulty with the concept of adaptation in biology is that, ".. it must be assumed that the partitioning of organisms into traits and the partitioning of the environment into problems has a real basis and is not simply the reification of intuitive human categories."  (Lewontin, 1984, p. 241)

Philosophical concern over the subjectivity of teleological explanation in biology often cuts even deeper for many theorists.  As expressed by Andrew Woodfield, this deeper concern is that, "We, the conceptualizing observers, project teleology on to an organism by subsuming it under the concept of a physical end."  (Woodfield, 1976, p. 30).  The concern here is that teleological ascriptions "project on to 'things' a property which the things do not possess." (Woodfield, 1976, p. 26).


This concern with the subjectivity of teleology is deeply related to the insight of Larry Wright, who has persuasively argued that teleological explanations are ascription explanations.  According to Wright, teleological ascriptions effectively explain what they characterize (Wright, 1972, p. 204).  In the formal context of evolutionary biology, this construal of teleological explanation generates two closely related questions.  The first question is raised by our need to account for the fact that a mere characterization can have full explanatory power.  The second question is raised by our need to account for fact that, at least within the in the context of evolutionary biology, teleological explanations may be further explained by the mechanism of natural selection.  Both questions reach to the heart of the mystery which is found in "the reasoning power which tells us plainly that most parts and organs are exquisitely adapted for certain purposes."  (Darwin, 1964, p. 453)

Let me suggest that the necessary beginnings of an adequate solution to this mystery is to be found in the schematic model provided by our own anticipated actions.  My argument is that teleological ascriptions -- like action ascriptions -- explain by assimilating observed patterns to the patterns provided by our own purposive behavior (see below).  Everyone, of course, admits that human beings routinely anticipate the future course of their own bodily movements.  Furthermore, even when it is hampered or obstructed the actual course of our own voluntary behavior remains focused upon anticipated bodily states.  It is this focus which gives character to our most basic voluntary actions.

In fact, actions which are characterized by their anticipated future states circumscribe a particular domain in kinesthetic space (see below).  They are patterned phenomena which are restricted to a certain range and are bounded within certain broad constraints (compare Wright, 1972, p. 205).  These behaviors are multiply realized patterns whose defining feature is a common focus upon anticipated but frequently unrealized future bodily states (see below).  Teleological ascriptions are explanatory because they posit phenomena which are multiply realized and circumscribed by a particular focus upon the future.  Frequently these posits are made contemporaneously with observation.  We are able to directly perceive teleological phenomenon as one instance of a whole array of multiply realized phenomena all of which are circumscribed by the same general focus upon the future.  In effect, our teleological observations identify a phenomenon as belonging to a pattern of a particular kind, and this identification is explanatory because it places the phenomenon within a class which is restricted within the bounds of the scope suggested by some future state (compare primitive action ascriptions as described below). 

Once a particular phenomenon has been placed within a teleological category it becomes familiar.  We are provided with a very general pattern of expectations which removes much of the surprise which might have been generated by the phenomenon under examination.  Although this identificary placement fails to provide us with unassailable reason to expect any one event or consequence to occur at some particular place and time, it does give us good reason to expect some general pattern of events or consequences to occur within some broad envelope of space and time.  In the case of teleological phenomena our expectation are fulfilled not by the occurrence of any single event but are instead fulfilled by the occurrence of a general pattern of events which matches the overall form of an anticipated abstract pattern.  This general pattern may be said to meet our most abstract expectations even as it fails to conform to our anticipations in the particular detail.  In fact, the pattern of expectations raised and assuaged by teleological identifications are very like those which grow from the mathematics of nonlinear dynamics. (Compare Hayek, 1952; Hayek, 1967;  Hayek, 1973; Hayek, 1978; Hayek, 1979; see also Campbell, 1984; Campbell, 1985; Prigogine, 1985; Crutchfield et al, 1986; Stone, 1989; Dennett, 1991)

We are able, for example, to model the abstract pattern of a dissipative chaotic system geometrically although any of the particular states of the system are not themselves predictable.  We are unable, that is, to predict any particular unknown state of a dissipative chaotic system by a calculation from some prior state of the system because any error of measurement will be magnified in our prediction of any future state (Stone, 1989, p. 126).  We are, however, able to account for the general pattern of the system based on the geometric form of our mathematical model.   As Mark Stone explains, ".. these paths [modeled by a non-linear equation] will converge to a recognizable geometric form, and for a given type of chaotic system the attractor will always be the same .. once a scientist has discovered the attractor of a chaotic system, then he has a model of the system, and that model will serve as an explanation."  (Stone, 1989, p. 129)

As Friedrich Hayek explains, models of this sort provide, ".. the explanation not of individual events but merely of the appearance of certain patterns or orders."  (Hayek, 1967, p. 40)  For example, Hayek points out that Darwin's theory of natural selection, ".. can explain and predict only kinds of phenomena, defined by very general characteristics:  the occurrence, not at a narrowly defined time and place but within a wide range, or changes of certain types; or rather the absence of other types of changes in the structure of the succeeding organisms."  (Hayek, 1967, p. 13; compare Scriven, 1959; Rosenberg, 1985, p. 65, p. 174)  In this manner, a model or theory of this general kind indicates only the general characteristics of events or a range of phenomena to be expected, rather than a singular event to be expected at a particular place and time (see Hayek, 1967, p. 11; Hayek, 1978, p. 33; compare Hempel, 1965, pp. 336-337; Salmon, 1990, p.9).

My argument is that teleological ascriptions are derived from the very paradigm of purposive phenomena:  human action.  No one doubts that we are each able to anticipate the future course of our own actions.  It must also be admitted that in many instances these anticipated bodily movements are not accompanied by any experience which could be identified as the reason or cause of these movements.  For example, we frequently bend a finger -- and fully anticipate doing so -- without contemporaneous experiencing any particular belief (e.g. that the bending of our finger will cause a gun to fire) or desire (e.g. that the man who stole our car should die) which could be provided as the reason or cause for our movements.  In these cases we simply instantiate our anticipated actions without regard to any consciously held 'motivating cause' like anger and without regard to the constraints of any consciously held belief about the effects of our actions upon the world beyond our body.

Simple bodily movements -- like the bending of a finger -- can be classified as primitive expressions of human agency which very often cannot be described or explained in terms of an agent's mental states -- if these are meant to refer to his past emotional state of mind or the consciously anticipated future consequences of his actions upon the world or upon his own emotional state of mind - i.e. to his motives.  In fact, it is my view that our intuitive recognition of primitive acts of human agency constitute paradigmatically explanatory identifications. 

For example, when we identify a bending of a finger as a voluntary activity we are identifying both the source of the action and its descriptive character in a fashion which is fully understandable to another human agent.  That is, because of our simple and direct ability identify another individual's bodily movement as one which was anticipated in advance by that agent we are able to specify both the nature of the action and the place from whence the action came.  In this case, the action came from a person like ourselves and the action was a bending of the finger with the same descriptive character as our own bending of the finger.     

Our own voluntary behavior, of course, constitutes paradigmatically teleological phenomena.  When our finger bends in the midst of our anticipating just that motion we can be said to have moved so on purpose. Yet the sort of primitive behavior represented by a bending of the finger for no discernable reason may frequently appear for some to fit somewhat awkwardly within our language for purposive phenomena.  This apparent awkwardness can be explained by the common metaphors which English speakers use to understand and characterize teleological phenomena.  George Lakoff and Mark Johnson, for example, argue that our concept of purpose is often understood and characterized in terms of our own purposive movements along a path from one point to another (Lakoff, 1987; Johnson, 1987).  According to this account, the experiential gestalt of our own physical movement from one location to another (e.g. by walking or crawling) constitutes a recurring image-schematic pattern consisting of three elements:  (1) a source, or starting point; (2) a goal, or end point; and (3) a sequence of contiguous locations connecting the source with the goal. 

This PATH schemata -- when it is joined to the metaphor STATES ARE LOCATIONS -- combines to produce the common metaphor PURPOSES ARE PHYSICAL GOALS.  This metaphor provides a conceptual structure which may be mapped or projected into deeply analogous situations.  As Johnson explains, ".. we are thus understanding very abstract purposes (such as writing a book, getting a Ph.D., finding happiness) in terms of the performance of various physical acts in reaching a spatial goal."  (Johnson, 1987, p. 114)  This projected understanding allows us to linguistically characterize these abstract purposes in terms of the PURPOSES ARE PHYSICAL GOALS metaphor.  For example,  Johnson suggests that the metaphor allows us to say such things as, "You have reached the midpoint of your flight training.  I've got quite a way to go before I get my Ph. D.  She's just starting out to make her fortune.  Jane was sidetracked in her search for self-understanding.  Follow me -- this is the path to genuine happiness."  (Johnson, 1987, p. 115)  

The metaphor PURPOSES ARE PHYSICAL GOALS can help to account for Charles Taylor's somewhat odd characterization of primitive agency as behavior which is goal-directed in the 'extended sense' (Taylor, 1964, p. 28).  For example, Taylor wishes to suggest that, "Implicit in our everyday notion of action is that of 'direction' to a goal or end . . We usually speak of a goal when there is a certain end-condition or change aimed at by an action ..".  Taylor, however, goes on to admit that, ".. there are many other actions which we shall also want to speak of as directed where there is no such end-condition by which they are characterized separately identifiable from the action itself . . Such are, for instance, dancing, walking, running, and so on.  Now we often want to speak of behavior classified in this way as directed, but what is being aimed at in these cases is not some end-condition but simply the emitting of the behaviour of the required type, and the end is not a result separately identifiable from the action but simply the action's having a certain form or fulfilling a certain description."  (Taylor, 1964, pp.  27-28)

It is the metaphor PURPOSES ARE PHYSICAL GOALS which has permitted many philosophers to erroneously insist that all human behavior must be determined by a particular set of beliefs and desires.  According to this view, purposive behavior can only be explained by citing the reasons which cause those behaviors to occur.  The intuition is that human actions are controlled by human purposes which are themselves simply reflective of human beliefs and desires about particular physicals ends and the alternative means for reaching those ends.  As an obvious result, goal-directed behavior is to be explained by citing the internal conscious states of an agent which cause a particular end to be either attempted or attained.    

However, the claim that goal-directed behavior is to be explained by citing the prior beliefs and desires which are the cause of that behavior cannot be sustained under an empiricist's universal regularity theory of causation.  Even David Hume himself admits that the relation between actions and motives does not conform to the standards of constant conjunction which an attribution of causation properly demands under the universal regularity account of causation (Hume, 1978, p. 661; compare Hume, 1955, p. 106; see also Ransom, mimeograph).  As Hume explains, "..the fear of death will always make a man go twenty paces out of his road; tho' it will not always make him do a bad action."  (Hume, 1978, p. 661)

It is worth noting that when we are in the process of carrying out an anticipated bodily movement we do not experience ourselves commanding specific muscles to contract to a given degree or in a particular sequence.  This observation reveals an important fact about the nature of our epistemic access to the character of our own bodily behavior.  It shows us that the anticipated general course of our own somatic behavior is essentially independent of our anticipation of the particular course of the individual bodily components which constitute those movements.  Notice also that our cognitive experience of the general character of an anticipated bodily motion is often expressible in much greater descriptive detailed than is our experience of the phenomenal particulars of any given instantiation of that motion. 

These observations taken together reveal the conceptual primacy of our anticipated general bodily movements over what we perceive in the actual course of our particular bodily movements.  They show us that the conceptual status of a particular bodily movement is established in advance by the conceptual status of the general bodily motion we anticipated prior to that movement.

In fact, the nature of primitive agency reveals a many-one relationship which exists been the many diverse instantiations of a particular movement and that one movement itself as it is anticipated.  Reversing this order, we may speak of the one-many relationship which exists between the one movement which is anticipated in advance of behavior and the many particular ways in which that one movement may be instantiated subsequent to its anticipation.      

We are now prepared to answer Wittgenstein's notorious question about the nature of human agency, "What is left over if I subtract the fact that my arm goes up from the fact that I raise my arm?"  (Wittgenstein, 1953, p. 161)  The simple answer to Wittgenstein's question can be expressed this way:  When I raise my arm I anticipate doing so.  Yet, when my arm goes up, only one of many possible ways of raising my arm is instantiated.  As a result, when I raise my arm as previously anticipated, what is left over in any particular instantiation of my arm going up is all of the other particular ways my arm could have gone up while still completely fulfilling my prior expectation that I would raise my arm. 

In fact, my ability to metaphorically characterize what I am aware of in the anticipation of raising my arm is descriptively more complete than is my awareness of the particular components experienced in any particular instance of my arm going up.  What is therefore left over in my phenomenal experience of the fact that my arm goes up (in terms of skin tension, muscle stress, etc) is the descriptive completeness directly understood in my prior awareness of an anticipated raising of my arm.

One of the principle virtues of the human paradigm model of teleology is to be found in its ability to account for our direct perception of teleological phenomena.  The model allows us to understand how the abstract patterns generated by our own behavior could serve as a lens for the direct perception of purposive phenomena.  As might be expected, this account of teleological perception is not so novel as to be utterly unusual.  Friedrich Hayek, for example, has suggested that the schematic patterns of our own movements provide master molds which act as perceptual templates in the observation of teleological phenomena.  Hayek goes on to argue that the ready-made patterns which develop in ordinary behavior and normal social life give us an innate sense for the physiognomy of biological events which serves to guide our observations of the environment.  (See Hayek, 1967, p. 18, p. 23, p. 43, pp. 51-57; compare Hayek, 1978; Wright, 1972; Wright, 1976; Ruse, 1981; Rosenberg, 1985, pp. 65-66; Edelman, 1987; Edelman, 1989)

The human paradigm model of teleology helps to account for the pattern and force of our purposive vocabulary when viewed in light of the linguistic work of George Lakoff and Mark Johnson (Lakoff and Johnson, 1980; Lakoff, 1987; Johnson, 1987; see also Hayek, 1967; Wright, 1976).  Lakoff and Johnson have developed an account of conceptual significance which locates much of our linguistic understanding within the structured gestalten which emerge from the dimensions of our own bodily experience.  They contend that the schematic patterns which are embodied in our own bodily movements provide the paradigmatic core for many of our most basic linguistic categories.  They also suggest that the holistic patterns which derive from our own motor activities serve as metaphorical devises which allow for the identification of resembling patterns across a variety of dimensions.

As an illustration of their view, Lakoff and Johnson have shown how our concept of causation finds its semantic home in the experiential gestalten which forms out our the motor activity of direct manipulation (see Lakoff and Johnson, 1980, pp. 69-76).  According to Lakoff and Johnson, our particular experiences of direct manipulation form a structured whole of properties which together constitute a more central or paradigmatic case of direct manipulation than do any particular experience of direct manipulation.  This structured gestalten serves in turn as the prototype upon which our linguistic category of causation is formed through an interactive process of pattern matching and social convention.  The concept of causation thus derives from the observed resemblance which individuals identify between our shared experiential gestalten of direct manipulation and a whole constellation of particular patterns perceived in the world.

In a similar fashion, I contend that our shared concept of purpose is founded upon the experiential gestalten which emerges from the schematic patterns of our own voluntary activities.  According to this account, the future focused many-one bodily schemata of anticipated behavior serves as an embodied metaphor or prototype which both structures our perceptions and shapes the boundaries of our shared linguistic category of purpose.  Teleological ascriptions are thus metaphorical extensions from the anticipated patterns of our own voluntary behavior which assert the membership of a particular phenomenon within a multiply realized pattern of phenomena which are commonly focused upon the future in some general manner.  (Compare Hayek, 1967; Hayek, 1978; Wright, 1976; Lakoff and Johnson, 1980; Rosenberg, 1985, pp. 65-66; Lakoff, 1987; Johnson, 1987). 


At this point in the paper it is worth taking a closer look at the way our shared patterns of human experience provide an intersubjective framework in which to explain the actions of ourselves and other individuals.  Everyone agrees that our anticipated future bodily states, when realized, frequently bring with them a sense of pleasure or the relief of some felt unease.  Often, but not always, these concomitant psychological states were expected as the generally reliable effects of our anticipated future bodily states.  When these psychological effects were consciously expected, we say that our bodily behaviors were motivated or inspired by these expectations, especially when these expectations carry with them strong feelings or emotions.  If we call any imagined future pleasure and feeling of relief an instance of desire, we might also call our psychologically charged expectations our motivating desires.

It should be noted, however, that our various motivating desires are not constantly conjoined with any particular consequent bodily motions.  It would therefore be a mistake to identify our motivating desires as causes -- if this term is meant to be used in a sense which is compatible with a universal regularity account of causation.  In fact, everyone agrees that we may choose to satisfy a motivating desire in a number of different ways or we may choose to leave the motivating desire unsatisfied -- the decision is up to us.  For example, when experiencing a momentary thirst for water we may raise a glass water to our lips, we may bend our head to a drinking fountain, or we may go on about our business and response not at all to this momentary thirst.  Now and again, of course, we may raise a glass of water to our lips without ever having experienced a prior thirst or any other particular motivating desire.  In fact, it should be obvious from this discussion just why the word 'motivation' serves as a ready term for non-constantly conjoined yet non-arbitrary general patterns which cannot be characterized statistically.  (See Lakoff, 1987; compare Hayek, 1967; Hayek, 1978; Campbell, 1984; Prigogine, 1985; Stone, 1989; see also Dennett, 1991).

It is nothing remarkable to suggest that when we observe another individual we immediately and without reflection place his actions within the usual pattern of our own behavior (compare Hayek, 1948).  For example, when we observe an individual raising a glass of water to his lips we directly imagine, comprehend and conclude that he has done so with anticipation of the future course of his bodily movements and in the context of a prior conscious thirst.  The placement of his action within this framework constitutes our explanation of his action.  This placement answers any number of different questions.  Among other things this placement tells us both why the action happened and how the action occurred.  Perhaps most importantly this placement tells us the origin of the action or where the action came from.  My claim here is that particular pattern placements of just this kind constitute the very paradigm of what we take an explanation to be. 

It is worth pointing to one item in particular which may be explained through reference to our shared general patterns of human experience.  Placement of a particular action within the framework of human actions allows us to explain why something is where it is or why it is there at all.  For example, our explanation of the action of taking a drink also provides us with an explanation for why a glass is where it happens to be.  We explain the placement of the glass of water on an individual's lip by the general pattern of behavior into which it fits.  Very often we shorten or paraphrase this explanation, perhaps somewhat misleadingly, by saying that the position of the glass of water is explained by the individual's prior thirst.  In fact, we frequently say that the glass of water is there for a reason -- which is an even more highly condensed reference to the larger framework of human action upon which it depends for its explanatory force.  This explanation may be particularly misleading to those who view causal ascription as the paradigm form of explanation and at the same time hope to maintain a constant conjunction construal of causation -- but of course a prior thirst is not constantly conjoined with a glass of water on the lips and a glass of water may end up upon ones lips for no prior reason that can be consciously identified. 

I would now like to introduce and discuss the common metaphor PURPOSES ARE BENEFITS which frequently shapes our conceptual understanding of both teleological and non-teleological phenomena.  The PURPOSES ARE BENEFITS metaphor can be seen in the work of many post-medieval European natural scientists and philosophers.  For example, the nineteenth century scientist David Brewster explained the purpose of the moon in terms of its beneficial effects for man.  According to Brewster, the purpose of the moon is to light the way of nocturnal travellers.  An implicit assumption of this explanation which falls out of the PURPOSES ARE BENEFITS metaphor is that the moon owes its place and being to the service it provides for man.  The moon benefits man by lighting the way for travellers in the night.  The fact that the moon was put there for the benefit of human beings explains why the moon is there.  In other words, Brewster, a Christian believer, implicitly infers from the PURPOSES ARE BENEFITS metaphor that the moon was put there by God for just this reason -- i.e. to light the way for man (compare Ruse, 1981, p. 89-91). 

The PURPOSES ARE BENEFITS metaphor can also be seen at work in the writings of Immanuel Kant.  It is just this metaphor which inspires Kant to characterize what are in fact the fortuitous benefits of nature as an expression of the deliberate intention of Providence to provide for the native inhabitants of the Arctic coast (Kant, 1970, p. 108).  This conclusion follows directly from Kant's metaphorical conception of purposes which suggests to him that the purposes of natural phenomena are those beneficial effects which they bestow upon man.  For example, under the sway of the PURPOSES ARE BENEFITS metaphor Kant is inspired to say that, ".. evidence of design in nature emerges .. when we realise that the shores of the Arctic Ocean are inhabited not only by fur-bearing animals, but also by seals, walruses and whales, whose flesh provides food and whose fat provides worth for the native inhabitants.  Nature's care arouses most admiration, however, by carrying driftwood to these treeless regions .. For if they did not have this material, the natives would not be able to construct either boats or weapons, or dwelling in which to live."  (Kant, 1970, p.110)

The common framework of human experience helps us to understand the internal structure and conceptual power of the PURPOSES ARE BENEFITS metaphor.  For example, a frequent result of the placement of a glass of water upon an individual's lips is the alleviation of the felt uneasiness of thirst.  We recognize and speak of this change of mental state as a good or benefit.  It is my claim that we can identify the origin of what I have called the PURPOSES ARE BENEFITS metaphor in the often beneficial consequences of our primitive purposive actions -- many of which are preformed in the context of a strong prior anticipation of these beneficial results.  My suggestion, therefore, is that the general patterns of human experience which consist of both anticipated bodily movements and expected physical and psychological outcomes provides not only the explanatory context in which to account for human action, it also provides the natural dimensions of experience which gives rise to the common metaphor PURPOSES ARE BENEFITS. 

I wish to show next how our awareness of the metaphor PURPOSES ARE BENEFITS can help us to account for the character of many of the most notorious anthropomorphic errors of the past.  Of course, metaphorical extensions by their very nature leave great room for error.  It is should not be surprising to find, therefore, that the PURPOSES ARE BENEFITS metaphor allows us to account for anthropomorphic error as an instance of unwarranted metaphorical over-extension.  My suggestion, then, is that a prime source of anthropomorphic error occurs when what are readily discovered to be simply fortuitous benefits for man are illegitimately placed within the purposive scheme which is structured by the PURPOSES ARE BENEFITS metaphor. 

A parallel example from the explanation of human behavior will help me to demonstrate this point.  As I argued earlier, we can explain why a glass of water is where it is -- on an individual's lips -- because we can fit this placement of the glass of water within the general pattern of human actions which consist of primitive purposive actions which are consciously focused upon generally reliable future benefits.  In cases of this sort we commonly say that the glass was put there for a reason.  The reason here is usually simply the benefit which is expected to follow from the placement. 

The common mistake which Brewster and Kant are led to by the PURPOSES ARE BENEFITS metaphor is to infer from a benefit for man back to a primitive purposive action.  The thinking here is that all benefits for man must conform to the general pattern of human behavior in which it can be explained why a thing is where it is by citing a prior expected beneficial outcome.  The driving intuition is that the beneficial placement of an item must be something which comes from a conscious agent with motivating desires.  Within this explanatory framework the particular placement or existence of items which benefit man is to be explained by citing the beneficial outcome which God or Providence had in mind for man.  In other words, the hidden assumption of Brewster and Kant is that a great many items were put where they are by God or Providence for just this reason -- i.e. because of their benefit for man. 

I now wish to argue that another important source of anthropomorphic error can be found in the metaphor PURPOSES ARE NONFORTUITOUS OCCURRENCES.  The inferential upshot of this metaphor can be identified in Aristotle's claim that, ".. things are either the result of coincidence or for an end".  (Aristotle, Physics, 199a)  My claim is that Aristotle's important distinction between chance causes and purposive causes is grounded upon the common metaphor PURPOSES ARE NONFORTUITOUS OCCURRENCES which finds its experiential basis in the universally witnessed distinction between the expected and unexpected outcomes of our primitive purposive actions.     

Evidence for the central importance within Aristotle's thought of the distinction between anticipated and unanticipated outcomes of human action can be identified in the paradigmatic examples which Aristotle chooses to provide of chance causes and purposive causes.  For example, in Aristotle's Nicomachean Ethics the paradigmatic case of an event caused by chance is the unexpected outcome of a human action -- e.g. finding a buried treasure unexpectedly.  (Aristotle, Nicomachean Ethics, 1112a).  In this example, it is the unexpected coincidence of the location of the treasure -- and not a conscious prior expectation -- which is seen by Aristotle as the cause of this particular result -- i.e. the finding of the treasure. 

By contrast, in Aristotle's Physics the very exemplar of a purposefully caused event is the anticipated outcome of a primitive action which is accompanied by a prior motivating desire -- e.g. walking about when motivated by a desire for good health (Aristotle, Physics, 195a -196a).  In this example it is the future state of health which was consciously expected in advance and not some coincidental event which is seen as the cause of this particular result -- i.e. the walking about.

Additional evidence supporting the claim that Aristotle derives his distinction between chance causes and purposive causes from the framework provided by the anticipated and unanticipated dimensions of human action can be found in Aristotle's classic example of a chance meeting in the marketplace.  In the Physics Aristotle describes a fictional scenario in which a citizen of the city unexpectedly receives money owed to him by another individual as the unintended consequence of his having gone to the market in the expectation of taking care of some entirely different matter of business.  In other words, his receiving money owed to him is the chance result of having gone to the market with another particular outcome in mind -- and not as a general rule (i.e. as a matter of habit with no particular outcome in mind) -- and not with the expectation of meeting a particular man and receiving money owed from him.  (Aristotle, Physics, 196a-197a). 

I would now like to display the inferential pattern which leads Aristotle from the PURPOSES ARE NONFORTUITOUS OCCURRENCES metaphor to the conclusion that both the growth of teeth and walking are things which come about for a given end.  For Aristotle, the end result of the growth of teeth and of walking -- i.e. mature teeth and good health -- are manifestly nonfortuitous occurrences -- they are aspects of the world which fit within a pattern and which do not appear as a simple matter of chance.  From these two examples -- walking and the growth of teeth -- Aristotle concludes that, "Events that are for the sake of something include whatever may be done as a result of thought or of nature."  (Aristotle, Physics, 196b)  In the particular case of mature teeth, the PURPOSES ARE NONFORTUITOUS OCCURRENCES metaphor allows Aristotle to reason that,  ".. things are either the result of coincidence or for an end, and [mature teeth] cannot be the result of coincidence or spontaneity, it follows that they must be for an end .. ".  (Aristotle, Physics, 199a)

This example leads Aristotle to conclude that aspects of the non-human world can be the result of purposive causes.  As Aristotle explains, ".. such things [e.g. teeth] are all due to nature .. Therefore action for an end is present in things which come to be and are by nature."  (Aristotle, Physics, 199a)  The growth of teeth is caused by the end of mature teeth, which is a part of nature.  For Aristotle, this implies that, "..Nature belongs to the class of causes which act for the sake of something.. " (Aristotle, Physics, 198b)    

Walking is another aspect of the world which Aristotle conceives of as coming to be as the result of a given end.  For example, Aristotle characterizes walking as a thing which is caused by health -- which is understood to be an end of walking.  When Aristotle speaks of health as the cause of walking he is speaking of health as a purposive cause (Aristotle, Physics, 196a).  Aristotle justifies this claim by citing the particular example of an everyday explanation of human behavior.  According to Aristotle, the following sort of response is commonly used to explain the walking behavior of an individual:  "'Why is he walking about?' we say.  'To be healthy', and, having said that, we think we have assigned the cause."  (Aristotle, Physics, 195a)


I want to argue now that the explanatory framework of human behavior can help to account for the truth contained in Michael Ruse's suggestion that, "the artifact model is the key to biological teleology" (Ruse, 1981, p. 93).  Very often, in fact, we find that the artifact model is applied by direct analogy to biological features.  Charles Darwin, for example, suggests that, "The tail of the giraffe looks like an artificially constructed fly-flapper ..".  (Darwin, 1964, p. 195)  A number of our most respected contemporary evolutionary biologists affirm that the model of human artifacts often serves as a valuable heuristic aid for the identification of biological adaptations.  As George Williams explains, "[It is] frequently helpful .. to recognize adaptation in organic systems that show a clear analogy with human implements."  (Williams, 1966, p. 10)

The heuristic value of the analogy to human artifacts for evolutionary biology is strong evidence for what I take to be Ruse's deeper insight that the model of human artifacts is an important window on the character of biological teleology.  My own suggestion is that biological features are placed within categories in a manner which is deeply analogous to the manner in which human artifacts are placed within categories.  Significantly, the close conceptual connection which exists between the categorization of biological features and human artifacts is especially evident in the important instance of artificial selection, which may serve here to illustrate the importance of usefulness to man as the prototypical feature of human artifacts -- and by extension also of teleological categories in biology (see below).  My suggestion, in other words, is that a characterization of the nature of artifact categorization is an important first step toward an adequate account of the explanatory significance of the human artifact model for biological teleology.    

Following Charles Darwin, we may distinguish between two types of artificial selection.  We may distinguish between:  (1) biological change which is effected unconsciously by man as the undesigned long-term result of isolated acts of selective breeding which lack any particular end in view; and (2) biological change which is effected deliberately by man as the intended or planned long-term result of methodical selection and breeding which is conducted with a distinct end in view.  The process of breeding and selection, however it is conducted, typically leads to the proliferation and improvement of biological features which Darwin alternatively describe as "good to the lights of man", as "valued" by man, as adapted to "man's wants or fancies", and as "manifestly useful or pleasing to man".  (Darwin, 1964, pp. 29-43) 

Let me suggest that the products of breeding and selection count as 'human artifacts' whether or not particular acts of selection and breeding lead to particular results which are useful and pleasing to man -- or whether or not particular results which are useful and pleasing to man were planned or anticipated in advance by the breeder.  In other words, as Charles Darwin describes, the process of 'artificial selection' operates regardless of whether or not biological changes have been produced as the intended result of human action -- and whether or not particular biological changes constitute results which in every instance are useful or pleasing to man.  My claim, however, is that the various categories of artificial selection (i.e. artificial biological features) are centered around specific criterions of human usefulness -- and that the various biological products of artificial selection, whether they are useful or not, are classified according to their analogical relation with or historical connection to paradigmatically useful biological features (compare Darwin, 1964, p. 32; Hull, 1985, 181-204).

I would like to contend, in other words, that human artifacts are placed within teleological categories as a result of their hoped for, anticipated or previously reliable useful effects upon ourselves -- or simply due to their deep structural analogy or historical connection with items which paradigmatically have such effects (compare Hayek, 1948; Hayek, 1979; Lorenz, 1977).  For example, a claw-shaped plastic hand, an electric rotatory brush, a spaghetti scoop, and a pencil may all count as back scratchers -- although only the first and second of these were designed and produced with the effect of relieving a back itch in mind, and although only the first and last have been used to produce such a result, and although the second is in fact mechanically incapable of producing this effect, and although the third has never been used to produce such relief -- and may or may not be capable of doing so.  The conceptual upshot of this sort of categorization is that each of these items has been vicariously assigned an imagined or hypothetical use based on either its sincerely anticipated or merely fancied potential effect -- i.e. that of relieving the discomfort of a back itch. 

Consider, in a similar fashion, all of the biological features which typically count as a pair of wings.  Think, for example, of the appendages of a bee in flight, of a hatchling sparrow in a nest, of an ostrich standing on the plains, or of a smashed butterfly on a windshield.  Wings, I would like to suggest, represent a classic biological category.  It should be clear from this list, however, that not all items which are readily identified as wings are capable of producing the paradigmatic effect we expect of wings -- i.e. flight.  Neither the wings of a hatchling sparrow, nor the wings of an ostrich, nor the wings of a smashed butterfly are capable of producing what we would call flight -- yet all are routinely classified as wings. 

My suggestion is that teleological categories in biology are framed around particular exemplars -- biological items -- which produce effects which are manifestly useful to organisms in a manner which is deeply analogous to the manner in which human teleological categories are framed around particular exemplars -- artifacts -- which produce effects which are manifestly useful to men.  Significantly, in either case a variety of items may fall within a teleological category because these items in some manner conform to the pattern of the exemplar of the teleological category in shape, placement, or history (etc.) although these items fail to produce the same manifestly useful effects.  For example, a back scratcher may fail to relieve itches of the back (e.g. a poorly designed electric back scratcher) or a pair of wings may fail to produce flight (e.g. the wings of a smashed butterfly). 

Let me suggest, in fact, that the explanatory status of biological features is similar to that of human artifacts.  The gross teleologicial features of biological entities are explanatory independently of an account of the underlying physical composition of the feature in the same way that the gross teleological features of human artifacts are explanatory independently of an account of the underlying physical composition of those artifacts.  As Daniel Dennet argues, ".. we can identify coin detectors as such -- we can figure out that this is the [supposed] competence that explains their existence -- long before we know how to explain, mechanically, how that competence is achieved (or better:  approximated).  (Dennett, 1987, p. 315 -- reading 'existence' here to mean 'why it is there'). 

The central place of manifestly useful biological items as the exemplar for teleological categories in evolutionary biology is well illustrated in the work of Charles Darwin.  On the biological category of wings, for example, Darwin is moved to exclaim that, "Nothing can be plainer than that wings are formed for flight, yet in how many insects do we see wings so reduced in size as to be utterly incapable of flight, and not rarely lying under wing-cases, firmly soldered together!"  (Darwin, 1964, p. 451)  This example points to the highly important tension which exists between the manifestly useful effects which define teleological categories and the often imperceptible yet easily imagined demands of natural selection.  As Darwin goes on to remark, ".. an organ useful under certain conditions, might become injurious under others, as with the wings of beetles living on small and exposed islands; and in this case natural selection would continue slowly to reduce the organ, until it was rendered harmless and rudimentary."  (Darwin, 1964, p. 454)


My suggestion is that teleological categories based on manifestly useful effects are what biologists call adaptations and that the continued action of the demands of natural selection are what biologists call the process of adaptation.  The problem created by this conceptual division, as Richard Lewontin explains, ".. is how species can be at all times both adapting and adapted."  (Lewontin, 1984, p. 238; compare Darwin, 1964, p. 95)  The central paradox of evolutionary biology is that the theory of natural selection explains the manifestly perceptible biological features of an organism -- i.e its gross teleological characteristics -- by positing the indiscernible action of a process which acts gradually upon imperceptibly small biological differences -- i.e. Darwin's 'slight variations' -- which influence long-term organism reproduction and survival rates.  (Compare Darwin, 1964, pp. 3-4, p. 6, pp. 33-35, p. 85, p. 91, p. 95, p. 460, p. 471; see also Maynard Smith, 1989, p. 4, and p. 37-38).  

My argument is that the explanatory gulf which exists between the manifestly teleological properties of organisms and the underlying process of natural selection accounts for why evolutionary biologists use the notion of selective competition and speak of the relative adaptedness of teleological features -- i.e. of the relative advantage of competing phenotypes which are recognized as serving the same purpose (compare Darwin, 1964, pp. 60-63, p. 433; Brandon, 1981, p. 97-98; Sober, 1984a, p. 42).  Charles Darwin, for example, speaks of the relative selective advantage enjoyed by larger antlers, thinker fir, the swiftest wolves, the largest glands, the most nectar, the swifter horses (artificial selection), and shorter or longer beaks (artificial selection).  (See Darwin, 1964, p. 90-92)  My claim is that we use the notion of competition when we do not know in advance the facts that determine the actions of competitors (compare Hayek, 1978, p. 179).  Let me suggest also that a direct consequence of the evolutionary biologist's need for the use of the notion of competition is that a specific hypothesis of relative adaptiveness often cannot be tested empirically (compare Hayek, 1978, p. 180; Rosenberg, 1985, pp. 166-172; Rosenberg, 1986, p. 198) 

What I will call the paradox of adaptation is manifested in especially illuminating terms by the conceptual and observational gulf which stands between the categorical clarity of Mendelian based population genetics and the infinitesimally small physical gradations which characterize most phenotypic differences -- a problem which is only compounded by the problematic many-many relationship which exists between genotype and phenotype (compare Lewontin, 1974; Hull, 1974, pp. 16-17; Kitcher, 1984; Rosenberg, 1985).  As Lewontin explains, "In contradiction to the discreteness of genotypic classes demanded by Mendelian analysis lies the quasi-continuous nature of the phenotypic differences that are the stuff of evolutionary changes."  (Lewontin, 1974, p. 21)  The problem of phenotypic gradation is found throughout the paleontological record.  As G. G. Simpson explains, speaking of a particular phenotypic difference in tooth character within the species which later became equine horses, "In these cases, and generally in similar paleontological data, the phenotypic expressions of mutations do not, even aside from the fact that they are smaller, have the clear-cut nature of such mutations as 'forked' in Drosophila or 'sinistral' in Partula.  Even when such characters first appear, frequently they cannot be classed as 'wholly present' and 'wholly absent,' but already show variation both in size and in extent; for instance, in the number and location of the teeth that show them."  (Simpson, 1944, pp. 60-61)

My claim is that the epistemic conditions which lead to the paradox of adaptation provide gross and insuperable barriers to reduction in biology.  The central difficulty is that the successful reduction of teleological features to physics and chemistry requires of the evolutionary biologist that he solve in the most minute physical and chemical detail what Alexander Rosenberg has described as, "the problem of rightly distinguishing functions from other beneficial and incidental effects of an item."  (Rosenberg, 1980, p. 75)  The problem here, as George Williams explains, is that we lack the kind of explicit criteria which would allow us to precisely distinguish adaptations from either fortuitous or merely incidental effects (see Williams, 1966, p. 8 and p. 15).  In order to acquire these criteria, however, we would need to be in an epistemic position sufficient to precisely state, as Richard Lewontin explains, "[w]hat are the physical schemata to which organisms are adapting and adapted."  (Lewontin, 1984, p. 247) 

Yet it is just these physical descriptions which are irredeemably beyond our ability to acquire in the needed detail due to our inability to adequately identify the appropriate dimensions of the environmental constraints which are shaping evolution through natural selection.  This result is a consequence of the fact that we have no independent means of ascertaining the precise selective dimensions of the environment which rivals the evidence seen in the direct effects of the environment upon a biological population itself.  In other words, because the features of organisms provide the best evidence of the dimensions of their own niches, we have no independent means of specifying the selective forces at work in the environment.  As Richard Lewontin explains, "There is no way to use adaptation as the central principle of evolution without recourse to a predetermination of the states of nature to which this adaptation occurs, yet there seems no way to choose these states of nature, except by reference to already existing organisms."  (Lewontin, 1984, p. 240) 

The demand for reference to the attributes of already existing organisms presents an insuperable barrier to the completed characterization of the physical states of a selective environment.  As Alexander Rosenberg explains, "Exactly how large the dimensionality of an evolutionary environment is we cannot tell, because we know too little about the structure and behavior of organisms, and about how external forces affect them."  (Rosenberg, 1983, p. 459)  This sort of difficulty is particular acute in the field of animal behavior, where experimental conditions in the laboratory often diverge radically from the selective environment which confronts the wild organism.  Indeed, Lewis Petrinovich, citing the work of Egon Brunswik, argues that, "systematic design [in behavioral psychology] almost inevitably involves the use of atypical backgrounds for the behaviors in question and embeds the behaviors in atypical contexts."  (Petrinovich, 1979, p. 374)  Petrinovich has exposed the pitfalls of such an approach through his own work on bird-song development in the white-crowned sparrow (see Petrinovich, 1989; Petrinovich, 1990).  Petrinovich's conclusion is that, "If we are interested in explaining how organisms behave in their environment .. we will have to study the environment as carefully as we do the organism."  (Petrinovich, 1979, p. 378)

It is easy to see how this requirement would represent an insuperable barrier to the fully completed physical or chemical characterization of either organisms or environments due to the reciprocal effects which each of these have upon the other.  As Lewontin explains, "The activity of all living forms transforms the external world in ways that both promote and inhibit the organism's life .. In this way the environment is a product of the organism, just as the organism is a product of the environment."  (Lewontin, 1984, p. 238)  My claim, then, is that if neither organism nor environment can be fully characterized independently of one another, then the dynamically relationship which exists between organism and environment will constitute an insuperable barrier to teleological reduction.

Recall again that an empiricist in the Humean tradition insists that for any legitimate teleological ascription in biology a specific physical or chemical reduction to 'causal laws' must 'in principle' be available to any scientist who manages to possess enough intellectual capacity.  For example, Ernest Nagel contends, ".. every statement about the subject matter of a teleological explanation can in principle be rendered in nonteleological language .. " (Nagel, 1962, p. 421)  In slightly different terms, Nagel alternatively insists that," .. every teleological explanation is translatable into an equivalent nonteleological one."  (Nagel, 1962, p. 421)  As was suggested earlier, for the Humean empiricist this means that it must be possible -- at least in theory  -- to reduce all of the teleological explanations in biology to the terms of some lower-level disjunction of universal laws and physical characterized items -- and as Kincaid has suggested this demand, "..minimally requires that someone or some group of people with sufficient computational ability could provide the relevant disjunct."  (Kincaid, 1988, p. 273).

I will now use the example of Ernest Nagel's work on 'directively organized systems' as a means for showing just why it is that our inability to provide a completed characterization of the adaptive environment constitutes an absolute theoretical barrier to teleological reduction in biology.  The idea of a directively organized system is the idea of a system which consists of several mutually independent component sub-systems each of which are linked together by a series of causal laws which serve to maintain the overall system at some particular goal state (compare Nagel, 1962;  see also Nagel, 1977; Rosenberg, 1985).  The difficulty for Nagel is to provide independent grounds and objective criteria for specifying any particular state as the actual goal state of the system.  As David Hull explains, "Nagel's definition of teleological system in terms of negative feedback depends on a prior knowledge of what counts as a goal-state."  (Hull, 1974, p. 109) 

The problem here, of course, is that the only rationale beyond our own gross and very immediate teleological perceptions for positing the existence of a specific goal-state for a biological system consists in the adaptive effects which the attainment and maintenance of such a state would have upon a biological organism.  As suggested earlier, however, the only way in which the precise adaptive effects of any biological items can be ascertained is through an adequate physical characterization of the adaptive environment.  The central problem for the reductionist, as Richard Lewontin explains, is that, " .. the external world can be divided up in a non-countable infinity of ways so that there is a noncountable infinity of conceivable ecological niches."  (Lewontin, 1984, p. 237)  But as we have already seen, because of the dynamically reciprocal relationship which exists between the characterization of organism and environment, a completed physical characterization of an adaptive ecological niche is not possible.  Without a completed physical characterization of the adaptive environment there can be no precise specification of the goal state of a system, and without a precise specification of the goal state of the system there can be no completed reduction of a bit of teleological phenomena to a specific disjunct of causal laws and physical descriptions.

Consider, for example, the case of human thermal regulation.  We might image that the process of regulating somatic relationships in the body and the process of regulating fat cell production are independent sub-systems which help to maintain a specific range of internal bodily temperatures which is the goal state of the directively organized system of thermal regulation.  The crucial question for the reductionist is this:  how we are to determine what the precise goal state of this system is?  My answer is that outside of the theory of natural selection we have no means for identifying such a norm other than our innate human ability to perceive teleological phenomena directly.  And yet, as I have already shown, within the theory of natural selection the precise specification of such a goal state is impossible.

We can see this through the example provided by Nagel of the goal state for human thermal regulation.  Nagel suggests that we characterize the goal state of human thermal regulation in terms of the internal temperature required to prevent a human being from sustaining fatal bodily injury.  As Nagel puts it, "Thus, the internal temperature of the body must remain fairly constant if it is not to be fatally injured.  In point of fact, the temperature of the normal human being varies during a day only from about 97.3 F to 99.1 F, and cannot fall much below 75 F or rise much above 110 F without permanent injury to the body."  (Nagel, 1962, p. 409)  Surely, however, for the purposes of natural selection the goal state of internal thermal regulation would be useless if it did not also provide an animal with the ability to reproduce -- yet we can imagine a particular organism which manages to maintain a bodily temperature which allows it to remain alive but without the ability to reproduce.

Despite their inability to establish physically precise specifications of the goal state for any particular system, biologists are still wont to speak of the 'given' needs of organisms and of the 'agreed upon' functions of particular biological features (compare Rosenberg, 1980, p. 61).  The intersubjective perception of goal targets among evolutionary biologists is particularly true in the case of animal behavior.  As Larry Wright explains, "In a large number of cases the goal-directedness of a bit of behavior is obvious on its face.  Many of our teleological judgments are as reliable and intersubjective as the run of normal perceptual judgments.  Occasionally there simply is no question about it:  the rabbit is fleeing, the cat stalking, the squirrel building a nest."  (Wright, 1972, p. 206)  Wright goes on to suggest that, ".. we should view with suspicion any analysis which contends that goal-directedness consists in a relationship among parameters of which we are quite usually ignorant in the contexts of these reliable judgments."  (Wright, 1972, p. 206)   

In fact, Nagel himself admits that his account of directly organized systems, ".. deals only with what is assumed to be the common distinctive structure of teleological systems .. " (Nagel, 1962, p. 418)  In other words, Nagel's own examples are not meant to be understood as a completed reduction of any particular teleological system -- but are meant instead as a blue-print for the logical structure of completed teleological reductions as they are effected by practicing biologists.  As Nagel explains, "This must suffice as an outline of the abstract structure of goal-directed or teleological systems.  The account given deliberately leaves undiscussed the detailed mechanisms involved in the operation of particular teleological systems; and it simply assumes that all such systems can in principle be analyzed into parts which are causally relevant to the maintenance of some feature in those systems, and which stand to each other and to environmental factors in determinate relations capable of being formulated as general laws."  (Nagel, 1962, p. 418)  Yet as I have attempted to show above, such an assumption is mistaken due to certain features inherent in the structure of biological explanation.


Another tack taken by some empiricist's in their attempt to motivate the demand for the in principle reducibility of teleological phenomena in biology is to imagine the process of natural selection from a position of complete omniscience -- i.e. from a God's eye point of view (compare Rosenberg, 1985; Rosenberg, 1986; Rosenberg, 1989).  Here the empiricist's account of teleological reduction to causal explanation typically takes the form of a two step procedure.  In the first step the empiricist identifies teleologically patterned phenomena as the sort of phenomena which are explained in terms of their effects.  In the second step the empiricist demonstrates how the effects of teleological phenomena can be tied to a set of physically characterized causal antecedents according to the causal laws of physics and chemistry.  With the successful completion of this second step, the reduction of teleologically explained biological phenomena to physics and chemistry is viewed as accomplished.

The full scope of these two steps is clarified somewhat by two other assumptions which inform the empiricist's reductive criterion for legitimate scientific explanation.  First, the empiricist assumes that every instance of a teleologically patterned biological phenomenon is one which can -- in principle -- be identified as the sort of phenomenon which finds its explanation in terms of its effects.  Second, the empiricist assumes that for every situation in which teleological phenomena have been identified the causal antecedents and the consequence classes of such a pattern can  -- in principle -- be given physical characterizations which may be linked by the causal laws of physics and chemistry.

Having posted these 'in principle' caveats, the empiricist typically goes on to claim that any biological phenomenon which can be given a legitimate teleological explanation can -- in principle -- be given a physical explanation.  Yet troubling questions about the epistemic status of the empiricist's reductive standard of scientific legitimacy are raised by the manner in which this claim is stated.  Why, for example, has the empiricist allowed for the mere 'in principle' reduction of teleological explanations to the causal laws of physics and chemistry?  Why, instead, has the empiricist not insisted upon the in fact reduction of all teleological phenomena to a nomically linked physical specification of their causal antecedents and consequence classes?  Why, indeed, doesn't the empiricist propose the actual completion of teleological reduction as his criteria for the scientific legitimacy of teleological explanations in biology?

A clear answer to these questions has recently been put forward by Alexander Rosenberg (see Rosenberg, 1985, pp.  61-62, p. 65, p. 118; Rosenberg, 1989, p. 254, pp. 257-262; compare also Kitcher, 1984, p. 350; Hayek, 1967, pp. 14-15, pp. 20-21, pp. 33-39).  Significantly, it is an answer thought to be fully compatible with traditional empiricist intuitions about the nature of scientific explanation.  On Rosenberg's account the empiricist's appeal to mere 'in principle' reduction is a simple concession to the numerical complexity of biological phenomena and to the computational limitations of the human brain.  For Rosenberg an empiricist's appeal to in principle reduction rather than to in fact reduction as the standard of scientific legitimacy in biology simply takes into consideration the manifest truth that while the computational capacities of the human brain are finite, the descriptive complexities of biological phenomena are -- for all practical purposes -- numerically infinite.

Let me be very clear about the precise force of Rosenberg's appeal to the cognitive limitations of the human brain and to the complexity of biological phenomena.  On the human side of the equation, Rosenberg argues that the cognitive limitations which are placed upon biological reduction can be accounted for as a simple product of the finite storage capacities and the limited computational powers of the human brain (see Rosenberg, 1989, pp. 260-260).  It is important to notice here that Rosenberg's account of the human cognitive limitations which restrict biological reduction does not appeal to the nature of human perceptual capacities and recognitional skills.  Neither does his account appeal to any inherent limitation which the physical make-up of the world and of our own sensory apparatus might place upon our perceptual access to the patterns which constitute the initial conditions and consequence classes which figure in scientific explanations.

In a similar vein, Rosenberg's notion of complexity in biology does not reflect the sort of mathematical, phenomenological or conceptual complexity which is today described by those studying the mathematical patterns of non-linear phenomena (compare Hayek, 1967; Campbell, 1984; Prigogine, 1985).  Rather, the notion of complexity which Rosenberg proposes as a restriction upon teleological reduction is a notion which simply reflects the immensely large number of different physical descriptions which would be required to complete the reduction of each and every teleological explanation to causal variables (see Rosenberg, 1985, p. 61 and p. 118; and compare Rosenberg, 1989, p. 254 and p. 262).

Rosenberg, however, does provide a rather elegant argument which explains just why biological phenomena should exhibit the sort of numerically rich descriptive complexity which -- for purely practical reasons -- prevents the completion of teleological reduction in biology.  Rosenberg's argument appeals to the fact that natural selection explains the persistence of particular biological structures by reference to a limited class of the effects of such structures.  Let me turn, then, to Rosenberg's explanation for the fact that biological phenomena is characterized by the sort of complexity which is naturally susceptible to teleological explanation.

Although nearly all writers on teleology have acknowledge the practical usefulness of teleological language to characterize the kind of phenomena found in the biological world, few physicalists have attempted to account for the natural applicability of this sort of expression within the biological sciences.  Among those who have, the majority have appealed to the fact that biological phenomena are naturally characterized by the display of what is almost universally recognized as a plasticity and persistence in their behavior (compare Nagel, 1977,  p. 272).

Empiricists have widely interpreted this plasticity and persistence as an indication that biological phenomena are formed out of a diversity of different physical structures and multiplicity of alternative causal pathways all of which are capable of producing or maintaining the very same state or effect (compare Nagel, 1977,  p. 272).  While many empiricists have felt satisfied merely with a causal interpretation of the readily apparent phenomenal features of biological phenomena, other empiricists have sought to account for the very existence of phenomena which are marked by such qualities.  The concern here is to explain the origin and persistence of phenomena marked by the behavioral features of plasticity and persistence while avoiding the sort of conceptual novelties which would take the physicalist outside of the domain of purely causal principles (compare Wright, 1968, p. 221).

Among the few empiricist's who have attempted to provide such an explanation, Alexander Rosenberg's selective account is by far the most comprehensive (see Rosenberg, 1985; Rosenberg, 1989).  Although Rosenberg's explicit concern is to provide a purely causal explanation for the numerical complexity of biological phenomena which makes the strict reduction of teleological phenomena to causal laws practically impossible, in the course of his account Rosenberg inadvertently provides us with a causal explanation for the origin and persistence of biological phenomena which naturally display the characteristics of teleological phenomena.

Rosenberg's unstated premise is that the mechanism of natural selections provides us with a purely causal explanation for the origin and persistence of phenomena which we naturally identify as teleological in character.  On this account, a multitude of diverse physical structures and alternative causal pathways all jointly share the sort of selectively significant common effects which lead to the further replication and persistence of these very same structures and pathways.  In fact, these structures and pathways are just the sort of phenomena which we invariably characterize as teleological phenomena.  As Rosenberg puts it, "Natural selection 'chooses' variants by some of their effects, those we identify as their functions."  (Rosenberg, 1989, p. 248)   

Rosenberg implicitly suggests that teleological characterizations are naturally applicable to the sort of phenomena which are selected for their common effects just when those common effects are produced by a multitude of different physical structure and causal pathways.  On this view, it is the operation of a process which selects for diverse structures with common effects which produces phenomena which can (and are) explained in terms of their effect. 

It is crucial here to distinguish between phenomena which are defined in terms of there effects and phenomena which are explained in terms of their effect.  For example, in the great majority of cases we do not identify the multiple causes of a single effect as teleological phenomena, nor, in the great majority of cases, do we identify the replication of the very same or identical physical structures or causal processes as teleological phenomena.  Thus we don't identify the common effects of acetic acid and hydrochloric acid on litmus paper as teleological phenomena, although we do define these chemicals as acids on account of their common effect (compare David Hull, 1982, p. 302). 

In a similar fashion, we do not identify the replication and persistence of the crystalline structure of a carbon diamond as teleological phenomena and nor do we identify the replication and persistence of a nuclear reaction as teleological phenomena.  Although the patterns of a nuclear reaction or of a carbon crystal persist through replication, these processes do not count as teleological phenomena because their causes and effects involve entities and processes which are structurally and causally identical.   

In effect, Rosenberg implicitly suggests that we are dealing with teleological phenomena when we are dealing with a diversity of different physical structures and causal pathways which consistently produce the sort of common effects which cause these various structures and pathways to persist through replication.  As Rosenberg puts it, "A given physical system may have indefinitely many effects on its environment.  Only a subset of these are actually identified functions -- i.e. selected for by the systems' environment."  (Rosenberg, 1989, p. 251) 

Rosenberg's strategy is to account not only for the natural applicability of teleological characterizations but also for the numerical complexity of biological phenomena by showing how functional language is used to identify a great diversity of structures and causal pathways all of which produce the same selective effects.  We need to take a somewhat closer look at this argument. 

Rosenberg's likens Darwinian evolution to a race in which duplicate prizes are awarded for first place in the' race' to be selected.  The contestants in this race are diverse physical structures and multiple causal pathways.  The duplicate 'prize' which is awarded to these structures and pathways is the opportunity to persist through replication.  Significantly, the criteria for winning this race is set by external causal forces which demand a specific common effect before the prize of replication is conferred.

Rosenberg's larger argument can be stated as the claim that the numerical complexity of teleological phenomena and the natural applicability of teleological language is the straightforward product of a causal process through which a variety of different physical structures and alternative causal pathways replicate and persist as a result of satisfying the same winning criterion.  Of course, this criterion is to be understood as a strictly causal criterion which consists in the production of a common effect which allows for the continued replication and persistence of these diverse structures and pathways.

Significantly, however, Rosenberg allows that a population of physically distinct structures and alternative causal pathways need not produce strictly identical common effects in order to be awarded a prize in the race for replication and persistence.  Rather, it is sufficient for the purposes of replication for a structure or pathway to finish a close second or third in the race of natural selection (see Rosenberg, 1989, p. 252).  Rosenberg explains that this anomalous effect is the direct result of the fact that non-selective accidents allow for the replication and persistence of diverse structures and pathways which produce less than optimally adaptive effects -- producing what evolutionary biologists call genetic drift.      

It seems, then, as if Rosenberg has given us not just one, but actually two different accounts of teleological phenomena.  On the first account, a function or goal-directed behavior can be identified as one of a diverse collection of different physical structures or causal pathways all of which are capable of producing causally identical effects of the sort which will allow for the persistent representation of these structures and process in a population (see Rosenberg, 1989, p. 248-249).  On the second account, a function or goal-directed behavior can be identified as one of a diverse collection of different physical structures or causal pathways all of which are merely similar rather than strictly identical in their effects, although all of which contribute to the replication and persistence of these very same structures (or at least resembling structures) in the members of a particular species (see Rosenberg, 1989. p. 252).

Under the first description, teleological phenomena can be characterized as structurally or causally diverse phenomena with precisely equivalent effects.  On this account, teleological phenomena form disjunctive classes of multiple causal pathways and physically distinct structures which produce strictly identical effects.  Although these are disjunctive classes, the disjuncts of such classes will be finite and well-defined due to the strict identity of their causal effects.

Rosenberg refers to such classes as "relatively restricted functions" and gives as a particular example the relatively restricted function 'codes for valine' which is satisfied by exactly three structurally distinct DNA sequences (see Rosenberg, 1989, p. 253).  Significantly, because the criteria for membership within the finite disjunctive class of a relatively restricted function is found in the precisely identical effects of diverse structures and pathways, the members of such a class may be independently identified or causally defined outside of the actual historical course of the evolutionary process. 

On the second account, Rosenberg broadens the criteria used to identify a function or a goal-directed behavior, extending to include structures and pathways which are only merely similar rather than strictly identical in their effects (see Rosenberg, 1989, pp. 252-253, especially where he says, "the systems are merely quite similar in respect of fulfilling one function."  [underlining added].). Although Rosenberg fails to indicate just what it is about the diverse effects of these structures and pathways which make them all in some sense similar, let me suggest that this similarity need consists in nothing more than the common effect these structures and pathways have in producing the same level of replication and persistence of entities sharing the same historical origins and which constitute parts of an evolving historical individual (compare Hull, 1989, pp. 126-127). 

I will refer to the members of such a class as "relatively unruly functions" and give as an example the relatively unruly function 'protects members of the species X from birds, or snakes, or the wind, or the sun, or will protect them from something else which is as yet unknown' which has been satisfied by the physical structures or causal pathways A, B, C , D, or will be satisfied by some other structure or pathway as yet unknown.  Like the members of a "relatively restricted function", the members of a "relatively unruly function" will constitute a disjunctive class.  But unlike the membership class of a "relatively restricted function" the membership class of a "relatively unruly function" not closed by any causal criterion and therefore is neither well-defined nor denumerably finite. 

This conclusion is a direct result of the phenomena of evolutionary drift which allows for the possibility of non-selective accidents in the course of evolution.  Significantly, then, because the criteria for membership within a relatively unruly function is only found within the actual course of evolution, the members of such a class cannot be independently identified or defined outside of the actual historical course of the evolutionary process (see below).  Rather, the membership of such class can only be discovered through an inspection of the actual outcome of the unfolding course of evolution (compare Stone, 1989, pp. 126-127).

As we have seen, Rosenberg wishes to show that the empiricist appeal to 'in principle' reduction is merely a practical concession to the numerical complexity which characterizes teleological phenomena.  His strategy has been to show how this numerical complexity has been the direct result of the process of natural selection.  Rosenberg's claim is that Darwinian evolution generates a diversity of different physical structures and causal pathways which are capable of producing the kind of common effects which contribute to the replication and persistence of these structures and pathways.  Rosenberg implicitly suggests that we identify these diverse structures and pathways as teleological phenomena just because they produce the kinds of effects which make it possible to explain these structures and pathways in terms of those effects.

On Rosenberg's account, then, the teleological phenomena we readily identify in the biological world cannot --in practice -- be reduced to a finite list of divergent structural and causal disjunctions as a simple result of the fact that natural selection generates so many different ways to produce the same selectively significant effects that men with finite brains are incapable of calculating and enumerating all of the different relevant structures and causal pathways. It should be obvious that Rosenberg's argument here depends only upon the existence of "relatively restricted functions" and could be made without reference to the class of "relatively unruly functions".  In fact, Rosenberg can frequently be read as presenting the argument in just this form.  (See, for example, Rosenberg, 1989, p. 252, where he says, "Natural selection thus makes functional equivalence-cum structural diversity the rule and not the exception .. This difference, resulting from the operation of selection for effects, explains why reduction goes smoothly in the physical sciences, but apparently not at all, in biology," (underlining added)). 

Indeed, the plausibility of Rosenberg's claim that teleological explanations are in principle reducible to the causal laws of physics and chemistry largely relies upon the purely causal identity conditions which are built into the category of "relatively restricted functions" which are defined by the fact that such function product causally identical effects.  By contrast, the category of "relatively unruly functions" includes a collection of structures and causal pathways none of which need produce identical effects, but which instead merely produce effects which are in some sense 'similar'. 

The challenge for Rosenberg and other empiricists is to show how the class of "relatively unruly functions" can in principle be reduced to the structures and causal pathways of physics and chemistry by showing us how the similarity relation which binds the members of any particular 'relatively unruly function" can be causally specified or defined.  The criteria for a successful demonstration that this relation and the classes it defines can in principle be reduced to the causal laws of physics and chemistry is of course the standard empiricist desiderata which demands that any scientifically reputable class or relation must represent properties or magnitudes which co-vary in a predictably linear fashion.  That is, the physicalist must in principle be able to define in advance the class of "relatively unruly functions" by "distinguishing exactly what is owing to chance, and what proceeds from causes" (Hume, 1985, p. 111).  And again, the physicalist must in principle be able to make such a distinction in advance of the appearance of the phenomena to be explained.  Let me now suggest why the physicalist will be unable to meet this challenge.

My argument can be illustrated through a parallel example from the domain of immunological theory.  The immune system is a selective system which produces hundreds of millions of antibodies each with its own unique bonding surface which adheres to a different range of invading antigen to a greater or lesser degree.  After adhering to an antigen, the cell bearing a particular antibody responds by replicating additional cells displaying more of the same antibody.  In fact, however, we don't know the particular function of any given antibody until we identify a particular antigen to which a cell bearing that antibody has had a response (compare Edelman, 1982, p. 32).  As Gerald Edelman explains, ".. hosts of antibodies exist that do not 'know' what they are for, like characters in a Pirandello play.  When a selective event occurs, however, they do 'know' what they are for .. In an immune system, a posteriori, an antibody stands for a particular foreign molecule.  But a priori this is not the case; when one looks at an antibody in the immune system, one does not know what it is for .. "  (Edelman, 1985, pp. 34-35)   

The a posteriori nature of functional attribution in immunology is particularly evident in the case of antibodies which bond to no currently known substance.  In effect, these antibodies have no currently known function.  Consider, however, an antibody of this sort which acts at some future date to remove an as yet unknown and therefore undeveloped hepten (i.e. man-made antigen which does not exist naturally in the environment) from the body.  The possibility of the existence of such a substance is demanded by the assumption the there is such as thing as the growth of human knowledge (compare Popper, 1957, p. vii).  As Karl Popper explains, "We cannot predict, by rational or scientific methods, the future growth of our scientific knowledge .. this refutes the possibility of predicting historical developments to the extent to which they may be influenced by the growth of our knowledge."  (Popper, 1957, p. vii)  It follows, then, that an antibody of this sort will acquire a purpose at some future date which no current scientist could predict in advance.

It is worth pointing out that the bonding surface of an antibody is typically degenerate -- meaning that the surface of the antibody is able to bond to more than one kind of antigen.  As a result -- in light of the possibility of the future development of currently unknown heptens -- there is no way to restrict the domain of the function of any particular antibody in the immune system based upon our current understanding of physical science.  In other words, the current state of physical science can neither circumscribe nor predict the precise delimit of future functional attributions for particular antibodies in the immune system -- nor can it circumscribe and predict the precise membership of the disjunctive set of physically defined microstructures which might mereologically constitute any given functional phenomena. 

Let me now suggest that Rosenberg's concession to the existence of the category of "relatively unruly functions" reflects our similar inability to draw a clear-cut physically defined line between genetic change which is a result of natural selection and genetic change which is a result of chance.  In order to show how we might be able to define this distinction physically, Rosenberg attempts to draw a clear-cut physical distinction between what he calls evolutionary and nonevolutionary forces (see Rosenberg, 1986).  According to Rosenberg evolutionary forces produce changes in gene-frequencies which represent a movement in the direction of greater adaptation to local conditions (see Rosenberg, 1986, p. 199).  On the other hand non-evolutionary forces are, " .. ones that operate irregularly, occasionally, in a way not normally correlated with changes in gene-frequencies."  (Rosenberg, 1986, p. 197)  As an empiricist who defends the in principle reducibility of biological phenomena to the causal laws of physics and chemistry, the challenge for Rosenberg is to explain how this distinction can be drawn in a fashion which is predicable in advance for any given case by a scientist with sufficiently large mental capacity -- a burden which I will show cannot be discharged. 

Let me first suggest that the fact that this distinction cannot be drawn for any given case in a fashion which is predicable in advance is both accommodated and masked within population genetics by the concept of genetic drift.  Genetic drift as it is conceived of by population biologists can be characterized as genetic change which is caused by sampling error in the effects of the environment upon on a biological population.  The distinction here is between differential reproduction which is a result of chance and differential reproduction which is a result of natural selection.  Rosenberg's claim is that the population biologist's notion of genetic drift is simply, ".. a place-holder for nonevolutionary forces that obstruct the effects of [natural] selection .. ".  (Rosenberg, 1986, p. 205)  Genetic drift is a place-holder, in other words, for irregular causal forces (sic) of which we are usually quite ignorant. 

If we understand biological adaptations in later generations to be the effects of natural selection, it should be clear on Rosenberg's account that nonevolutionary forces can be understood to be exceptional and unknown selective forces which do not produce adaptive effects in later generations (compare Rosenberg, 1986, p. 198).  Rosenberg therefore insists that under a proper interpretation, "drift will refer to entirely deterministic, but utterly non-evolutionary factors, which .. through our ignorance we must treat probabilistically."  (Rosenberg, 1986, p. 197) 

In effect, according to this account, the possibility of the in principle reduction of biological features requires that a scientist with a sufficiently large brain could at once eliminate the population biologists notion of genetic drift in favor of a purely causal distinction between evolutionary and non-evolutionary selective factors and at the same time gain access to all of the causal determinants which contribute to genetic change. 

To show that this is not possible we need only look at the hidden assumptions which must be smuggled illegitimately into any hypothetical example of a nonevolutionary selective factor.  Proposed examples of nonevolutionary selective factors can be found in the work of both Alexander Rosenberg and Elliott Sober.  Rosenberg provides the following example:  " .. through a freakish never-to-be repeated wind conditions, the two tallest trees on which the most attractive vegetation for giraffes grows, become so twisted that they accidentally trapped the heads of most of the tallest giraffes, causing their necks to break, and leaving them ensnared."  (Rosenberg, 1986, p. 197)  Rosenberg adds, "In this case, the change in gene frequencies may be said to be due to .. nonevolutionary factors .. ".  (Rosenberg, 1986, p. 197)

The illegitimate assumption here is that the non-repeatability of this selective event can be predicted in advance by a scientist with sufficiently large intellectual resources.  To begin with the example provides no principled reason for ruling out the repeated occurrence of relevantly similar wind conditions.  In fact, the example actually relies upon the gross relative similarities between giraffes, trees, vegetation for purposes of identification -- which effectively provides the necessary room for the repeatability of a selective event with relevantly similar features.  This reliance upon the gross relative similarities between features is a concession to the relative inaccessibility of the initial conditions for any selective event beyond the gross similarities which are recognizable within the crude range of direct human perception.  In fact, any attempt to gain closer and more precise access to the initial conditions for any selective event would disturb just those conditions which are relevant to the interaction between the environment and a biological population for natural selection (compare Hull, 1973, p. 137; see also Stone, 1989).

The central difficulty for those who wish to draw an a priori distinction between evolutionary and nonevolutionary selective events is that is impossible in advance to provide at the same time clear criteria for identifying an exact similarity between selective events and contemporaneously providing clear criteria for distinguishing between selective events which are repeatable and selective events which are not repeatable.  The issue here can be illustrated with an example from Elliott Sober.  Sober asks us to:  "Imagine identical twins walking on a mountaintop.  One of them is struck by lightning and dies; the other is unscathed and goes on to reproduce."  (Sober, 1984a, p. 118)  Sober adds, "It is intuitive to ascribe this difference in reproductive success to 'chance'."  (Sober, 1984a, p. 118)  This would seem to be a clear case of nonevolutionary selection under Rosenberg's definition in that the selective event seems to be a one time occurrence.  Sober also asks us in addition, however, to, " .. [s]uppose [that] the twins differed in some way that would explain the fact that one lived and the other died.  Perhaps the one struck by lightning was walking on a slightly higher path and this more exposed position slightly increased the chance of a lighting strike.  (Sober, 1984a, p. 118)  Sober then asks, "Would this fact force us to redescribe the event as involving natural selection?"  (Sober, 1984a, p. 118)  In fact, this looks very close to qualifying as an instance of evolutionary selection under Rosenberg's definition in that the selective event seems to be clearly repeatable in a fashion which is linked to a difference between the twins.  Sober then asks us to imagine that, ".. the first twin usually walked on the highest, most exposed, part of the mountain, but the second twin tended to walk along the less lofty paths .. ".  (Sober, 1984a, p. 119)  If this phenotypic difference between the twins could be linked to an inadvertent genotypic difference between the twins, then the selective event which is represented by the lightning strike would seem clearly to be a potentially repeatable and therefore evolutionary selective event. 

The necessary hidden assumptions necessary to rule out this possibility are the assumptions that no such genotypic differences exist and that deadly lightning strikes on the highest most exposed part of the mountain are not repeatable events.  Because of the inherent limits noted just above upon discovery of the initial conditions necessary to verify these assumptions no such possibility can be ruled out in advance.  We must instead wait for the unfolding of events in order to discover whether or not what we will later identify as evolutionary selection events do in fact eventuate.  In effect, then, we cannot distinguish in advance between adaptive and nonadaptive biological features.  As Gerald Edelman explains, "All selective theories stress that there is variation whose origins are not directly influenced by a posteriori events.  For example, in evolution, mutations occur spontaneously but the significance of particular gene mutations cannot be accessed until some natural selective condition acts upon the phenotype."  (Edelman, 1982, p. 31)  And as we have just seen, we cannot distinguish between natural selective events and nonnatural selective events outside of the unfolding of historical events. 

Often, in fact, we cannot identify outside of a wide range what exactly these selective events are.  For example, evolutionary biologists have long speculated over exactly what purpose was served by the bony plates of the stegosaurus.  What sort of selective events might account for the existence of this distinctive biological feature?  Did they serve to frighten off predators, of did they serve to facilitate courtship recognition, or did they serve to regulate body temperature? (Compare Ruse, 1981).  Without a clear choice between these open possibilities the empiricist would seem to be at a loss to reduce these biological features to any particular set of physical disjuncts.

My conclusion based on all of the above difficulties for reduction is that within evolutionary biology we must be satisfied with the prediction and explanation of the abstract characteristics of a general pattern of events and we can never hope for anything more precise than a multiply-realized and often open-ended or incomplete how-possible causal explanation of a particular event appropriately identified as falling within the domain of biology.

Greg Ransom
Dept. of Philosophy
February, 1992


Baker, Lynne Rudder.  1987.  Saving Belief,  (Princeton:  Princeton U. Press).
Bartley, W. W. III.  1987.  Evolutionary Epistemology, Theory of Rationality, and the Sociology of Knowledge,  (LaSalle:  Open Court).
Beauchamp, Tom; and Rosenberg, Alexander.  1981.  Hume and the Problem of Causation,  (Oxford:  U. of Oxford Press).
Bradie, Michael.  1986.  “Assessing Evolutionary Epistemology”, Biology & Philosophy, 1, 401-459.
Brandon, Robert.  1981.  “Biological Teleology:  Questions and Explanations”, Studies in the History and Philosophy of Science, 12, 91-105.
Brandon, Robert.  1990.  Adaptation and Environment, (Princeton:  U. of Princeton Press).
Campbell, David.  1984.  “Introduction to Nonlinera Phenomena”, in D. Campbell, ed. 1984.
Campbell, David.  1984.  Order and Chaos.
Campbell, Donald.  1985.  “Pattern Matching as an Essential of Distal Knowing”, in H. Kornblith, ed. Naturalizing Epistemology, (Cambridge:  MIT Press).
Carnap, Rudolf.  1956.  Meaning and Necessity, (Chicago:  U. of Chicago Press).
Carnap, Rudolf.  1966.  The Philosophy of Science, (New York:  Basic Books).
Darwin, Charles.  1964.  On the Origin of Species, (Cambridge:  Harvard U. Press).
Clark, A. J.  1983.  “Meaning and Evolutionary Epistemology”, Theoria, 49, 23-31.
Clark, A. J.  1984.  “Evolutionary Epistemology and Ontological Realism”, Phil. Q., 34, 482-490.
Clark, A. J.  1986.  “Evolutionary Epistemology and the Scientific Method”, Philosophica, 37, 151-162.
Crutchfield, James; Farmer, Doyne; Packard, Norman; and Shaw, Richard.  1985.  “Chaos”, Sci. Am., 255, 46-54.
Dennett, Daniel.  1991.  “Real Patterns”, J. of Phil., 27-51.
Edelman, Gerald.  1982.  “Through a Computer Darkly:  Group Selection and Higher Brain Function”, in Bul. of the Am. Acad.  of Arts & Sci., 36, 18-49.
Edelman, Gerald.  1985.  “Neural Darwinism:  Population Thinking and Higher Brain Fuction”, in M. Shafto, ed. How We Know, (San Francisco; Harper & Row).
Edelman, Gerald.  1987.  Neural Darwinism. (New York:  Basic Books).
Edelman, Gerald.  1989.  The Remembered Present.  (New York:  Basic Books).
Hanson, Norwood.  1958.  Patterns of Discovery, (Cambridge U. Press).
Hayek, Friedrich.  1948.  Individualism and Economic Order, (Chicago:  U. of Chicago Press).
Hayek, Friedrich.  1952.  The Sensory Order.  (Chicago:  U. of Chicago Press).
Hayek, Friedrich.  1967.  Studies in Philosophy, Politics, and Economics, (Chicago:  U. of Chicago Press).
Hayek, Friedrich.  1973.  Law, Legislation, and Liberty, Vol. 1, (Chicago:  U. of Chicago Press).
Hayek, Friedrich.  1978.  New Studies in Philosophy, Politics, Economics, and the History of Ideas, (Chicago:  U. of Chicago Press).
Hayek, Friedrich.  1979.  The Counter-Revolution of Science, (Indianapolis:  Liberty Press).
Hempel, Carl.  1965.  Aspects of Scientific Explanation, (New York:  The Free Press).
Hempel, Carl.  1966.  Philosophy of Natural Sciece (Englewood Cliffs:  Prentice-Hall).
Hull, David.  1974.  Philosophy of Biological Sciece, (Englewood Cliffs:  Prentice-Hall).
Hull, David.  1989.  “Central Subjects and Historical Narratives”, in D. Hull, 1989.
Hull, David.  1989.   The Metaphysics of Evolution, (Albany:  SUNY Press).
Hume, David.  1955.  An Inquiry Concerning Human Understanding, (Indeianapolis:  Bobbs-Merrill).
Hume, David.  1978.  A Treatise of Human Nature, (Oxford:  U. of Oxford Press).
Hume, David.  1985.  “Of the Rise and Progress of the Arts and Sciences”, in E. Miller, ed., Essays Moral, Political, And Literary, (Indianapolis:  Liberty Press).
Kim, Jaegwon.  1978.  “Superveniance and Nomological Incommensurables”, Am. Phil. Q., 15, 149-156.
Kincaid, Harold.  1987.  “Supervenience Doesn’t Entail Reducibility”, S. J. of Phil., 25, 343-355.
Kincaid, Harold.  1988.  “Supervenience and Explanation”, Synthese, 77, 251-281.
Kincaid, Harold.  1990.  “Molecular Biology and the Unity of Sciece”, Phil. of Sci., 57, 575-593.
Kornblith, Hilary, ed.  1985.  Naturalizing Epistemology, (Cambridge:  MIT Press).
Kuhn, Thomas.  1970.  The Structure of Scientific Revolution, 2nd Ed., (Chicago:  U. of Chicago Press).
Lewontin, Richard.  1974.  The Genetic Basis of Evolutionary Change, (New York:  Columbia U. Press).
Lewontin, Richard.  1984.  “Adaptation”, in E. Sober, ed.  Conceptual Issues in Evolutionary Biology, (Cambridge:  MIT Press).
Lakoff, George.  1987.  Women, Fire, and Dangerous Things, (Chicago:  U. of Chicago Press).
Lakoff, George; and Johnson, Mark.  1980.  Metaphors We Live By, (Chicago:  U. of Chicago Press).
Lorenz, Konrad.  1977.  Behind the Mirror, (New York:  Harcourt Brace Jovanovich).
Maynard Smith, John.  1989.  Evolutionary Genetics, (Oxford:  Oxford U. Press).
Millikan, Ruth.  1984.  Language, Thought, and other Biological Categories, (Cambridge:  MIT Press).
Mises, Ludwig.  1978.  The Ultimate Foundations of Economic Science, (Kansas City:  Sheed Andrews & McMeel).
Nagel, Ernest.  1977.  “Teleology Revisited”, J. of Phil., 261-301.
Nagel, Ernest.  1979.  The Structure of Science, (Indianapolis:  Hackett).
Newton-Smith.  1981.  The Rationality of Science, (London:  Routledge & Kegan Paul).
Petrinovich, Lewis.  1979.  “Probabilistic Functionalism”, Am. Psych., 34, 373-390.
Petrinovich, Lewis.  1989.  “Representative Design and the Quality of Generalization”, in L. Poon, D. Rubin, B. Wilson, eds., Everday Cognition in Adulthood and Late Life, (New York:  Cambridge U. Press).
 Petrinovich, Lewis.  1990.  “Avian Song Development:  Methodological and Conceptual Issues”, in D. Dewsbury, ed., Contemporary Issues in Comparative Psychology, (Sunderlund:  Sinauer Assocs.).
Popper, Karl.  1957.  The Poverty of Historicism, (New York:  Harper & Row).
Popper, Karl.  1962.  Conjectures and Refutations, (London:  Routledge & Kegan Paul).
Popper, Karl.  1968.  The Logic of Scientific Discovery, (New York:  Harper & Row).
Popper, Karl.  1979.  Objective Knowledge, 2nd ed., (Oxford:  Oxford U. Press).
Popper, Karl.  1983.  Realism and the Aim of Science, (Totowa:  Rowman & Littlefield).
Prigogine, Ilya.  1985.  “New Perspectives on Complexity” in S. Aida, et al, eds., The Science and Praxis of Complexity.
Putnam, Hhilary.  1983.  Realism and Reason, (Cambridge:  Cambridge U. Press).
Quine, Willard.  1953.  From a Logical Point of View,  (Cambridge:  Harvard U. Press).
Quine, Willard.  1960.  Word and Object, (Cambridge:  MIT Press).
Quine, Willard.  1990.  Pursuit of Truth, (Cambridge:  Harvard U. Press).
Rorty, Richard.  1979.  Philosophy and the Mirror Nature, (Princeton:  U. of Princeton Press).
Rosenberg, Alexander.  1980.  “Causation and Teleology in Contemporary Philosophy of Science”, in Contemporary Philosophy, Vol. 2, 51-86.
Rosenberg, Alexander.  1982.  Sociobiology and the Preemption of Social Science, (Baltimore:  Johns Hopkins U. Press).
Rosenberg, Alexander.  1983.  “Fitness”, J. of Phil., 457-473.
Rosenberg, Alexander.  1984.  “The Supervenience of Biological Concepts”, in E. Sober, ed.,  Conceptual Issues in Evolutionary Biology.
Rosenberg, Alexander.  1985.  The Structure of Biological Science, (Cambridge:  Cambridge U. Press).
Rosenberg, Alexander.  1986.  “Is the Theory of Natural Selection a Statistical Theory?”, Can. J. of Phil.  Supp. Vol. 14, 187-207.
Rosenberg, Alexander.  1989.  “From Reductionism to Instrumentalism?”, in M. Ruse, ed., What the Philosophy of Biology Is, (Dordrecht:  Kluwer Academic).
Ruse, Michael.  1986.  Taking Darwin Seriously, (New York:  Basil Blackwell).
Ruse, Michael, ed., What the Philosophy of Biology Is, (Dordrecht:  Kluwer Academic).
Salmon, Wesley.  1990.  Four Decades of Scientific Explanation, (Minneapolis:  U. of Minnesota Press).
Schaffner, Kenneth.  1969.  “The Watson-Crick Model and Reduction”, Brit. J. Phil. Sci., 20, 325-348.
Scheffler, Israel.  1963.  The Anatomy of Inquiry.  (New York:  Knopf).
Scriven, Michael.  1959.  “Explanation and Prediction in Evolutionary Theory”, Science, 130, 477-482.
Simpson, George.  1944.  Tempo and Mode in Evolution, (Columbia:  Columbia U. Press).
Sober, Elliot.  1984a.  The Nature of Selection, (Cambridge:  MIT Press).
Sober, Elliot.  1984b. Conceptual Issues in Evolutionary Biology, (Cambridg:  MIT Press).
Stone, Mark.  “Chaos, Prediction and Laplacean Determinism”, Am. Phil. Q., 26, 123-131.
Suppe, Frederick.  1989.  The Semantic Conception of Theories and Scientific Realism, (Chicago:  U. of Illinois Press).
Taylor, Charles.  1964.  The Explanation of Behavior, (London:  Routledge & Kegan Paul).
Van Fraasen, Bas.  1989.  Laws and Symmetry, (Oxford:  Oxford U. Press).
Williams,  George.  1966.  Adaptation and Natural Selection, (Princeton:  U. of Princeton Press).
Wittgenstein, Ludwig.  1953.  Philosophical Investigations, (New York:  Macmillan).
Wittgenstein, Ludwig.  1958.  The Blue and the Brown Books, (New York:  Harper & Row).
Wright, Larry.  1972.  “Explanation and Teleology”, Phil. of Sci., 39, 204-218.
Wright, Larry.  1976.  Teleological Explanation, (Berkeley:  U. of California Press).
Woodfield, Andrew.  1976.  Teleology.  (Cambridge:  Cambridge U. Press).