Final causation is out of favour in modern scientific circles, as it is incompatible with a "scientific world-view that countenances only efficient causation" (Buller, 1999, p. 6, italics mine). Accordingly, those who wish to do away with teleological talk in science have two choices: eliminate the very concept of teleology, or reduce it to something more scientifically respectable.
Eliminativism, as we shall see below, is not a viable option. Teleology is real. Proponents of final causation, such as Cameron (2000, 2004), have therefore focused their attack on reductionism, arguing that attempts to reduce teleological properties to structural properties or historical properties fail to account for their goal-directedness and biological normativity.
My own intermediate position, which I shall elaborate in section 1.B, is that (i) the structural and/or historical properties of some part of an organism can indeed determine whether it is goal-directed and subject to biological norms, but (ii) it is impossible in principle to describe the formal and historical properties of these parts without employing some teleological language, hence (iii) the attempt to reduce teleology to some other explanatory "dimension" of life is doomed to failure. On point (i), I part company with Cameron. In this section, I shall limit myself to briefly cataloguing the formidable (and as yet unsolved) philosophical problems that any reductionist program needs to address. (Cameron (2000, pp. 168-213) contains a very comprehensive discussion of these problems.)
The failure of eliminativism
The "eliminativist" option appears unworkable: as biologist Karen Neander points out, "the apparent explanatory power of teleological explanations which appeal to biological functions is quite robust" (1991, p. 127). (Cameron (2000, pp. 171, 219-220) lists citations from several other scientists who uphold the relevance of teleological explanations.) Neander calls teleology the "conceptual glue" of biology and notes that it would be "hard to exaggerate" the concept's importance to biology (1995, p. 227).
Walrus tusks provide a good illustration of this point. The question of what they are for is surely a meaningful one, and it is certainly legitimate, from a scientific standpoint, to investigate the many functions they serve (for fighting, especially during the mating season; for protection against predators; as anchors, when hauling their bodies out of the water; as pickaxes to cut a path through the ice; and for anchoring themselves on the ocean bottom while digging for clams).
The inadequacy of eliminativism leaves us with the option of assimilating teleology to something else - in other words, "to analyze the concepts of being a goal and being goal-directed so that the analytic versions of these concepts ... contain no undefined teleological notions" (Nagel, 1979, p. 291). The two most popular proposals for realising this project are the systems approach to teleology and the etiological account of functions, both of which are critically surveyed by Cameron (2000, pp. 174-213; 2004).
The systems approach to teleology
The systems approach endeavours to explain teleology as a structural feature of certain complex systems. A well-known problem with this approach is that the same structural features can be found in systems which nobody would describe as teleological. Bedau (1992) argues that even if we make these features more robust, non-teleological counter-examples with the same structural features can still be found. In other words, "a system's [intrinsic] causal dynamics do not by themselves determine whether the system is genuinely goal-directed" (Bedau, 1992, p. 265).
It is, of course, possible that living things do in fact have unique structural features, which have not yet been properly described or enumerated by scientists. I shall return to this point in section 1.B.1. However, I would like to point out that even if (as I suggest below) the structural features of a system can determine whether it is goal-directed, that does not establish the success of the reductionist program. In order for the program to succeed, it needs to be shown that these structural features can be described without any reference to their goal-directed properties. I shall argue below that it is precisely at this point that the reductionist program falters.
The etiological account of functions
Most contemporary scientists attempt to account for the occurrence of biological functions, in terms of an etiological account. The key idea here is that items with proper functions must have a history of selection. An item's past history (which made it biologically advantageous) explains its present use: "the proper function of a trait is to do whatever it was selected for" (Neander, 1991, p. 124). For instance, "grasping objects was what ... the opposable thumb was selected for, and that is why it is the function of your thumb to help you grasp objects" (Neander 1991b, p. 130).
Varner (1998) provides a clear, attractive summary of the etiological account, based on the work of other authors in the field:
X is a biological function of S (some organ or subsystem) in O (some organism) if and only if:(a) X is a consequence of O's having S and(b) O has S because achieving X was adaptive for O's ancestors (1998, p. 67).
In appealing to history and natural selection to explain teleology, the etiological account draws upon two of the five "dimensions" I described in section 1.A.2 - the temporal and efficient causal "dimensions". The etiological account can therefore be considered as a reductionist explanation of teleology.
One advantage of explaining functions on an etiological account is that it can easily explain the functionality of an organ in awkward cases where the creature possessing it is unable for accidental reasons (e.g. injury or sterility) to benefit from it. Thus even if an animal is blind or unable to reproduce, its eyes can be said to possess a function such as sight, because its ancestors enjoyed a selective advantage by being able to see. However, a straightforward non-etiological explanation is also possible: the animal's eyes can be said to possess a function because eyes in other (healthy) members of its species have a function.
If the etiological account is right, the teleological language we use when talking about living things can be grounded in their history of natural selection. However, there are formidable problems with the etiological account, which are described in detail by Cameron (2000, pp. 176-213).
First, the etiological account undermines the authority of biologists whose specialty is not evolution (e.g. physiologists) to attribute biological functions to organs, without reference to their evolutionary past. We accept the statement that the heart's function is to pump blood, simply because physiology tells us so; but if the etiological account is correct, any attribution of function made by a physiologist is in principle revisable, if contrary evidence from evolutionary history comes to light.
The fact that it is possible for scientists to deduce the functionality of an organ or biological subsystem in a living creature, without looking at its ancestors, is a major stumbling block for the etiological account. As Richard Dawkins puts it:
[A]ny engineer can recognize an object that has been designed, even poorly designed, for a purpose, and he can usually work out what that purpose is just by looking at the structure of the object (1986, p. 21, italics mine).
It is certainly true that scientists do sometimes talk about the original function of a structure that presently has none (e.g. vestigial legs in snakes) and that they also distinguish between the original function and the present function of an organ or biological subsystem. For instance, the opiates in an animal's body not only have the biological function of pain relief, but also of attacking bacteria and sending signals to the immune system, which appears to have been their original function (Stefano, Salzet and Fricchione, 1998). In these cases, the etiological account performs a valuable service of rendering intelligible our attribution of original functions to biological structures. But even here, the attribution of an original function is made by comparing the structure or subsystem with equivalents in present-day organisms, as the organs of ancestral forms are seldom preserved. This is precisely what Stefano, Salzet and Fricchione (1998) did: they examined the biochemistry of living invertebrates.
A second problem with the etiological account is that it flouts what appears to be a reasonable categorial constraint on proper theoretical definitions of concepts: namely, that a proper definition of an existing concept should not attempt to place it in a new ontological category. We cannot re-define the number 2 as a color, because category constraints prevent us from doing so. Likewise, it is philosophically illegitimate to attempt to re-define functions as historical properties.
This point is borne out by counter-examples which appear to show that a history of natural selection is neither necessary nor sufficient for the possession of a biological function. It is not necessary, because we can conceive of cases where we would still confidently ascribe functions to a creature's organs, even if they had no evolutionary history: if an instant lion appeared in our midst, we would have no hesitation in saying that the function of its heart was to pump blood. Nor is it sufficient: there are actual cases of entities (clay crystallites) possessing complex structures that undergo a process of natural selection (Cameron, 2000, p. 205). No-one has suggested that these inorganic entities are alive, and in any case, their parts appear to have no function: they are not for the sake of anything.
Although Cameron's real-life counter-examples to the notion that an organ's having history of natural selection is sufficient for its possessing a natural function are persuasive, his argument against the necessity of a history of natural selection is that it relies heavily on thought experiments relating to situations which, while logically possible, may or may not be really possible. In the Introduction to my thesis, I cautioned against reliance upon arguments of this kind.
Fortunately, there are better cases at hand. Varner (1998, p. 67) himself acknowledges that his definition of "biological function" fails to confer functionality on the first organism to possess an adaptive mutation, such as a photosensitive spot, or "proto-eye" - a difficulty to which Varner responds by suggesting that new traits acquire biological functions only via subsequent selective pressure, which presumably means that the first organism on earth had no functions!
Artificial selection poses an even weightier problem. For it is by no means clear why a function that arose through artificial rather than natural selection should not be considered biological. For instance, there are some species of eyeless fish whose ancestors possessed a sense of sight but who have since lost it because they live in underwater caves where having eyes is actually a biological disadvantage, as the eyes are liable to bacterial infection if injured from the fish's occasional collisions with cave walls. The DNA in these fish still contains the genetic program for constructing eyes, but it has been switched off. If future scientists succeeded in switching on this eye-making program, and released a group of genetically engineered fish into an environment where sight was an advantage, we would surely say that the function of their descendants' eyes was to see, even though their ability to do so was the product of artificial rather than natural selection.
Despite these troubling exceptions, we should not overlook the fact that the etiological account encapsulates a powerful insight: that all present-day functions of organs or biological subsystems in living things arose because they were advantageous to their ancestors. The etiological account is therefore universal in scope, if we leave aside genetically engineered organisms.
A final problem with the etiological account, according to Cameron, is that it is incapable of generating biological norms, and explaining how organs can malfunction, or fail to do what they are supposed to do. The fact that a part is adapted and selected by Nature for doing something does not endow the part with a biological function, any more than my artificial selection of a trait in a lineage of laboratory organisms would give that trait a function:
There is neither purpose nor function to natural selection, nor resulting from natural (or artificial) selection. What is selected is simply selected, and what the selected does it merely does (2004, p. 5).
I am not persuaded by Cameron's argument on this point. There is an important difference between natural and artificial selection: in the former case, the trait is selected because its properties make it intrinsically advantageous in the organism's biological habitat; whereas in the latter case, the trait is selected for non-biological reasons. In the natural case, there is a real sense in which the trait per se promotes the organism's survival, given its biological environment: the trait itself explains the selective advantage of organisms possessing it. This warrants our making the biologically normative statements that (i) organisms of the same species, living in the kind of environment where the trait confers a selective advantage, should have the trait; (ii) the trait has the function of doing whatever it does to improve the organism's chances of survival.
Cameron also argues that the fact that an organ has worked in the past does not generate any biological norms about how it should work now or in the future. But if an organ "works" over a period of millions of years (under a certain range of environmental conditions), then it is sensible to assume that there is some law of nature that explains why it works. If there is such a law, then our reliance on it to hold in the future (under the same range of conditions) is no more problematic than our reliance on any other law - e.g. gravity. The biologically normative content of the law reflects the underlying notion that biological norms are simply those kinds of behaviour that inherently tend to promote an organism's survival.
However, even if etiology can generate biological norms, a number of outstanding problem cases remain. As Cameron (2004) points out, etiology per se is neither necessary nor sufficient to yield such biological norms: an instant lion's heart - or, if we employ my less controversial example, the eye of a genetically modified cave fish - could malfunction, and there is nothing that clay crystallites are supposed to do, despite their history of selection.
Thus far, we have focused on the philosophical inadequacies of the etiological account of functions. However, even if the foregoing problems could be resolved, I contend that there is a more fundamental problem with the etiological account: it fails to reduce teleology to the dimensions of time and efficient causality, as it set out to do.
It is important to distinguish here between the following questions that can be legitimately asked regarding any biological organ or subsystem S:
1. How did S arise?2. How did S become prevalent?
3. Why did S become prevalent?
4. Why does S persist?
5. What is S for?
The Darwinian answer to the first question is: random variation. (Hence the question: "Why did S arise?" is meaningless.) The etiological account claims to provide an answer to the remaining four questions, reducing the fourth and fifth questions to the third. What I would like to point out here is that even the third question cannot be answered without reference to (a) the organism as a whole, and (b) advantages enjoyed by previously existing organisms subsequent to the acquisition of the trait. I shall employ the example of the opposable thumb to illustrate my point.
1. How did the opposable thumb arise? (Answer: by a process of random mutation.)
2. How did the opposable thumb become prevalent? (Answer: it conferred a selective advantage on individuals possessing it.)
3. Why did the opposable thumb become prevalent? (Answer: because individuals born with this trait were subsequently able to grasp objects once they became old enough to fend for themselves, giving them a selective advantage.)
4. Why does the opposable thumb persist? (Answer: because individuals born with this trait are able to grasp objects once they are mature enough to fend for themselves, giving them a selective advantage.)
5. What is the opposable thumb for? (Answer: it is for grasping objects.)
Of the four questions which the etiological account claims to answer, only question 2 is purely historical: any existing organ or subsystem with a biological function became prevalent in the general population because its possessors lived longer and left more descendants than their rivals. However, question 2 cannot tell us what the function of the relevant organ or subsystem is; all it tells us is that the organ or subsystem must have some function, as it is biologically advantageous. To discover what the function is, we need to answer question 3, at the very least.
But even if we reduce questions 4 and 5 to question 3, we cannot dispense with the holistic reference to the organism, nor can we avoid referring to "future" states or abilities of the organism, subsequent to the acquisition of the adaptive trait (e.g. its ability to grasp objects when it is old enough to fend for itself). To explain why the trait is present, we have to "look ahead".
I conclude that etiology, despite its promising insights, is incapable of reducing the teleological aspect of biological functions to a more scientifically "respectable" combination of efficient causation and history.
Can etiology be used to explain what it means to be alive?
According to Varner's (1998) bold proposal, an etiological account of functions not only explains functionality, but can also be used to explain what it means for an organism to be alive.
Varner makes a sharp distinction between biological functions and an organism's built-in goals, or end-states. What is good for an organism cannot be adequately defined in terms of its end-states, as artifacts - such as a Patriot missile - may also have "built-in goals". (I shall revisit Varner's example of the missile in section 1.B.5.) Instead, Varner proposes that "biological functions, rather than goals or end-states" are required to "draw a sharp distinction between all artifacts, on the other hand, and all living organisms on the other" (1998, p. 67).
According to Varner, the crucial difference between organisms and artifacts is that organisms evolve:
One thing that distinguishes organisms from artifacts is that the former but not the latter are the result of natural selection (1998, p. 69).
Because all living things evolve, it is appropriate to ascribe biological functions to their organs or subsystems (e.g. the function of eyes is to enable their possessor to see). Artifacts do not possess functions; they merely have needs. Varner thus distinguishes life from non-life in terms of functionality, rather than finality - exactly the opposite of the position that I am arguing for.
However, Varner's claim that artifacts lack built-in functionality is empirically false, as it rests on the assumption that natural selection is a feature of natural systems only. As we have already seen, natural selection can occur in abstract computational systems too (Lenski, 2003). Additionally, one can envisage concrete, physical artifacts which undergo natural selection - for instance, a von Neumann probe, described in a thought experiment by the physicist Frank Tipler (1982). As Tipler conceived it, such a probe would travel around the Galaxy, guided by an in-built computer "capable of self-replication and capable ... of constructing anything for which it has plans, using the raw materials available in the solar system it is aimed at" (1982, p. 34). If the von Neumann probe also had a modifiable program, which occasionally (when the program mutated) acquired new functionalities that help them to make more replicants of themselves, then these functionalities would be subject to natural selection in their different planetary environments.
Varner's proposal that we can account for the distinction between life and non-life in etiological terms thus appears to be at odds with the available evidence.
I conclude that teleology remains an irreducible, ontologically primitive fact of life.