Functions. The recognition of supervenient properties is the most common way of describing the failure of reductionism, [1] and functional properties are the example that has forced philosophers of science to recognize that some properties are supervenient. The main problem is that they may be realized by indefinitely many different and seemingly unconnected sets of physical traits, as illustrated by such artifacts as clocks. As we have noted, clocks may be realized by objects whose physical structures range from machines worn on the wrist to tree rings, sun dials, and the amount of radioactive decay. Artifact are a special case, which is one sense are not so problematic, because we know that they depend on the intentions of subjective beings. In another sense, they are more problematic, because it requires the reduction of intentions. However, functional properties also play an enormous role in biological, where they pose a similar problem. For example, hearts are mechanisms for circulating energy in multicellular organisms, but they cannot easily be picked out by their physical properties, because they vary from simple gastrovascular cavities to elaborate circulatory systems with arteries and veins involving one or more hearts of various kinds.
There is no general agreement about the significance of the existence of supervenient properties. At one extreme, they are considered a way of defending physicalism against the claims that there are processes that cannot be explained in terms of the laws of physics. At the other extreme, they have attracted other philosophers of science toward emergentism, is the sense of the belief that there are laws in less general branches of science that cannot be reduced to physics.
Though supervenience theorists generally deny that upper level laws mentioning supervenient properties are irreducible, supervenient properties do entail a kind of law that is not reducible.
Supervenience theorists insist that the causal connections in which supervenient properties may be involved can always be explained by the physical causes that are responsible for the regularity in that case, though, of course, those physical causes vary with the kind of physical properties that realized the supervenient property in that case. They are right to deny that there is any need irreducible causal laws (that is, laws that can be used to explain events by efficient causes). No one believe that clocks or hearts require anything but physical laws for their operation.
Even so, however, the reduction of the less general regularity to the laws of physics is not complete, because the physical explanation of what happens in each instance of a supervenient property does not explain the indefinitely large variety of different sets of physical traits that may realize the supervenient property. In other words, the grouping of those cases in such a way that they all have the same supervenient property is itself a regularity that has not been explained. If supervenient properties are anything more than purely subjective projections onto the world, then the fact that such physically diverse objects can be grouped together in describing upper level regularities is something that needs to be explained in the end. That regularity may not be a law of nature in the sense of a law of nature that supports an efficient cause explanation (according to the deductive-nomological model). But it does imply the existence of an order of some kind about the world, and that order cannot be reduced to the laws of physics and conditions described in physical terms.
In the case of functional properties, furthermore, the prime example of supervenient properties, there is even more reason to suspect that there are irreducible laws of nature, because functional properties are typically used to give functional explanations. It is not just that certain organs in multicellular bodies are all have the function of circulating energy to all parts of the body, but that the existence of such organs seems to be explained by that function. But if functions are causes that can explain the traits that have them, it involves causal connections like those in efficient causes. The function is a different event or condition from the trait it would explain, just as the efficient cause is different from its effect.
It is certainly not like the connection between an ontological cause and its effect. The function is not constituted by the trait described physically. If it were, there would be nothing supervenient about the functional property. Instead, traits are said to “realize” the function, because there are many different ways that functional properties can be realized. But that makes it even more mysterious how the function can be said to explain the trait, since the same function can be served by physically different traits.
Furthermore, since a functional explanation explains the trait by the function, it would not help to discover that the trait constitutes (or realizes) the function, because the function must be prior to the trait to “cause” it. And that would require explaining where the function comes from or how it could make material objects have the physical properties that would constitute them.
The prior issue is, therefore, whether functional explanations in biology are valid and, if so, how. Though most philosophers are inclined to believe that they are valid in some sense, there has been no generally accepted defense of their validity. The received view is that they are really just disguised historical explanations of a contingent process of selection, which do not justify prediction of the traits that will evolve.[2] But if evolution is a global regularity in a world of matter and space in time, there is a sense in which they are valid explanations.
There is, of course, no question of functional explanations being valid, if that means that the function is a substance that acts on matter to give it the trait, that is, to give it the physical properties that enable it to serve the function. That is the kind of causal connection entailed by Aristotelian teleology, or what is called “final causation.” Aristotle believed that having an essential form would make natural change take place in the particular substance for the sake of an end, final cause, or telos, which is said to be good for substances of its natural kind. Naturalists have long since recognized that there are no essential forms in the natural world that work in the way Aristotle supposed. That is entailed by materialism about the natural world, which has prevailed since the beginning of modern science, and essential forms acting as final causes are not among the substances assumed by spatiomaterialism.
The validity of functional explanations. The validity of functional explanations in biology is, however, entailed by ontological philosophy, and the way in which their validity is explained, confirms their validity in a far stronger sense than is currently recognized. Functions do cause the traits that serve them, and if evolution is due to reproductive causation, functions explain why organisms have the traits they do in a way that makes it possible, in principle, to predict that the traits will evolve.
That is not to say that every physical property of the traits is predictable. The traits usually involve some physical properties that could be otherwise. But enough of the physical properties of the traits are determined by their functions that they can be recognized by their physical properties in the organisms.
Evolution is due to reproductive causation. That is, evolution is a global regularity that is explained ontologically as the kind of change that is constituted by reproductive cycles and the wholeness of space. The reproductive cycles are material structures of a certain kind using the available free energy to go through cycles in which they both reproduce and do non-reproductive work that controls conditions that affect their reproduction. The regularity about change in the region over time includes , as we have seen, both a gradual change during each stage in the direction of maximum holistic power for organisms of their kind (or their natural perfection) and a series of evolutionary stages in the direction of the natural perfection of life itself.
Since reproducing organisms impose natural selection on themselves (by the scarcity caused by generations of reproduction in space), what is regular about change in the region over time is that every possible increase in the power of the reproducing organisms is necessarily made actual as it becomes possible. Each random variation of their structures that is acquired because it controls some condition affecting its reproduction is a trait. Its function is to control the relevant condition. And since the conditions that it is possible for random variations on evolving organisms to control are “in the cards,” so to speak, they can, in principle, be used to predict the traits that will evolve.
Likewise, for the revolutionary episodes. The higher levels of part-whole complexity in the structures of the reproducing organisms that can be tried out at each stage of evolution depend on the natures of the reproducing organisms that already exist, because they must originate as a radical random variations on existing structures. And whether they can control some relevant condition that was previously out of reach depends on the nature of the region where conditions affect their reproduction. That is also “in the cards,” so to speak, and since both the possibility and the functionality can be known, the stages of evolution are, in principle, predicable. Thus, once again, even higher levels of structure in reproducing organisms can be explained by the function that it is possible for such structures to serve.
Actual predictions of the new traits that occur in gradual evolution by their possible functions would require the capacity to imagine every possible random variation and to see what condition those secondary effects would control, and that is usually not possible. Thus, it is only after the change has occurred that we are usually in a position to see which possible function was responsible for the trait's evolution. But in the case of revolutionary evolution, it is easier to see the possible functions of new kinds of primary structures, and that is the kind of functional explanation that was used to trace the course of evolution in the previous section.
This ontological explanation of evolution as a global regularity entails, in other words, a necessity about the kind of change that takes place over time in the region of space. It is a kind of regularity that makes prediction possible, in principle.
This is formally similar to the explanation of dispositions and ordinary causal connections between events described in the last chapter, for those regularities were also global regularities explained by matter and space as ontological causes. In dispositions, the event ordinarily called the “cause” is typically the way free energy is supplied, and the irreversible change that takes place is the effect.
In this case, however, reproductive causation necessarily makes every possible increase in the power of primary structures actual, and given the meaning that "function" has ontologically, that means that it necessarily makes every functional trait that is possible actual. Possible functions are, therefore, the cause of the evolution of certain kinds of secondary effects in much the same sense that compressing and releasing an elastic object causes it to spring back or putting a sugar cube in water causes it to dissolve. The evolutionary changes that make it possible for the random variations on reproducing organisms being tried out to be functional in a new way are what causes that trait to evolve.
Though functional explanation are valid, the functions are not essential forms with causal powers, as Aristotle assumed. In Aristotelian teleology, functions are assumed as a basic principle (if not substance) of the ontology, and thus, their causal powers are not explained, but merely assumed. But in evolution by reproductive causation, the ontological causes are the kinds of space and matter that exist in a world like ours, for they are the ultimate ontological causes of reproductive global regularities.
The ontological reducibility of functional properties. The predictability of traits by their functions should remove any doubts about the reducibility of functions or functional properties to the ontology of naturalism. Doubts about their reducibility come from the understanding that contemporary Darwinists have of the causes of evolutionary change. They are, as pointed out in the explanation of reproductive global regularities, accidentalists. They think of natural selection as being imposed on living organisms from outside by unpredictable changes in their environment, and they worry about the availability of random variations to meet the new conditions in the best possible way. For them, in Kauffman's (1993) words, evolution merely "cobbles together jury-rigged contraptions.”
This view of evolution is another example of the effect of overlooking the wholeness of space as a cause of regularities about change over time, for instead of seeing evolution as the way that reproductive cycles add up in space over time, it sees evolution as driven by an externally imposed natural selection. Thus, it seems to contemporary Darwinists that different traits might have served the same functions. Since that means that there is no necessary connection between functions and the traits that serve them, functions are said to be "supervenient properties" relative to the physical nature of the traits that have them.
But if there is a necessary connection between the functions and the traits that serve them, as implied by their status as consequences of spatiomaterialism, then functional properties are, in principle, reducible to the ontology of naturalism. This is to reduce functional properties to our ontology in much the same way that we reduced dispositional properties, except that the relevant global regularity depends on reproductive causation, rather than structural causation.
It is the progressiveness of evolutionary change that entails the validity of functional explanations and the ontological reducibility of functional properties. From the beginning, I have described evolutionary change as change in the direction of natural perfection, and I have distinguished various kinds of natural perfection: the natural perfection of the organisms at each stage, the natural perfection of their combination in the ecology, and the natural perfection of life in the series of stages of evolution. Even evolutionary change itself has a kind of natural perfection about it because of the way that what happens at each moment contributes to the progress.
The direction of evolutionary change was called “natural perfection,” because it always involves a maximum holistic power and that is the kind of part-whole relation that is optimal in a spatiomaterial world. It is “natural” perfection, because it is the kind of perfection that is appropriate in a natural world.
Though it depends on the thermodynamic flow of matter from forms of free energy to energy bound as evenly distributed heat, nothing can structure thermodynamic order except material structures, and reproductive causation is making the most of structural causation by shaping reproducing organisms to be as powerful as possible in using the available free energy to control conditions in the world. To be sure, until the evolution of reason, organisms acquire only those powers that control conditions that affect their own reproduction. But that is simply what is required for structural causes that are maximally powerful to exist in a world of matter and space in time. No structural causes, regardless how powerful, would last very long, if they did not use their power to ensure their own existence. Organisms do that in a way that makes them as powerful as possible, and rational beings do that because it is good.
The natural perfection produced by reproductive causation made it possible to explain goodness as contributing to natural perfection. Each part of such optimal part-whole relations makes a necessary contribution to its maximum holistic power, and thus, each is good in the sense of contributing to the natural perfection of the whole of which it is part. And as we have seen, this explanation is a definition of “good” that vindicates all our deepest and firmly held convictions about what is good and bad (and about what is right and wrong).
By this definition, goodness and perfection are related to one another as the property of the part is to the property of the whole in the products of reproductive causation. When the whole is perfect, all the parts are as good as they can be, and when all the parts are as good as they can be, the whole is perfect.
Moreover, it follow that the function that each non-reproductive structural effect has is good for the organism of which it is part, that each kind of organism is good for the ecology of which it is part, and that each level of organization in the structures of organisms that comes to exist with new stages of evolution are good for what exists in the whole region in which evolutionary change is happening. Ultimately, therefore, there is one whole on the planet (or planetary system) to whose perfection all the good parts make a necessary contribution.
To be functional is, therefore, to be good. Since their functions explain the traits that evolve, what explains the traits that organisms have is their goodness. The goodness of the random variations is what explains why they are naturally selected. Likewise, since what explain each new stage of evolution is the functionality of its higher level of part-whole complexity, what explains each new stage of evolution is its goodness. The goodness of the higher level of organization is what explains why it is naturally selected. This connection to the nature of goodness is another way of saying that evolution is progressive.
[1] John Post (1987) re-defines "physicalism" as a kind of materialist ontology that rejects reductionism in favor of what are, in effect, supervenient properties, and he goes so far as to take that anti-reductionism as the main reason for accepting it.
[2]A good statement of the etiological theory is given by Larry Wright (1973, 1976), but see also Michael Ruse (1973). For a criticism of the etiological theory and a defense of what they call the "propensity theory", see Bigelow and Pargetter (1987). Karen Neander (1991) defends the etiological theory against their criticisms, but in a way that is not very convincing, at least, not to me.