Wednesday, January 5, 2011

A Reconsideration of Teleology

In recent months I've been forming some serious doubts about the idea that teleology is something real in nature. Part of the reason for this has been the consideration that objective teleology seems only to be a feature of designed systems, whereas natural systems have evolved randomly. At a deeper level, however, my issue has been with the idea of essential properties, and it's correlative theory of "natures" (both being at least necessary, if not sufficient, conditions for a teleological system). By nature I mean that thing, or set of things, that demarcates one species or genera from another. As Aristotle put it in the Metaphysics, a thing has a nature insofar as it has a source of change intrinsic to itself (thus, for example, a tree has a nature because it has a natural end-directed growth process that arises from it's own, internal, teleological structure). However, I am beginning to reconsider the idea of natures as a viable philosophical option.

One of my main objections to the idea of essential properties has been, roughly, this: There is no non-arbitrary way to differentiate between essential, and non-essential (or "accidential"), properties. And that, because of this, the idea of the "essence" of a thing is a human idea with no objective reality. The arbitrariness of the distinction can be seen by way of the following line of reasoning: If there are essential properties, then any object x that lacked any one of its essential properties would cease to be x. This, however, has the absurd consequence that it is only adequately functioning members of a species that actually are members of that species. For example, if it is essential to a human being that he be a rational animal, then if he ceases to be rational (say, is born with major brain defects, or has an ailment such as schizophrenia), then he has, by definition, ceased to be a human being. This, though, is an unwanted conclusion, and thus the essentialist must give up the idea of essential properties.

However, an obvious reply is quick at hand. It is not those properties that an object x actually manifests that constitute its essence, but rather it is those properties that x would manifest if x were functioning properly. So, in the case of our irrational human being, the property "rational" is essential to him in the sense that if he had not been born with such a brain defect, then he would, under normal circumstances, behave in such a way that we would normatively call "rational" (he would avoid walking off cliffs, make plans for the future, and the like). This, however, seems prima facie unsatisfactory. For if the essence of a thing were a set of potential properties (made potential by, say, the thing's genetic structure), then how could you demarcate one potential property as "essential" from any other? To make this thought more clear, consider the fact that not only does my genetic structure give me the potential to manifest, say, sight (call this an essential property), but it also gives me the potential to manifest arbitrary, what the essentialist wants to call "accidental", properties such as "enjoying roast beef". So looking to the set of potential properties given in my genetic structure seems not be a useful criterion for demarcation either.

It is because of these and related issues that I began to doubt the idea of essences, or natures, in the natural world. The idea seemed incoherent at worst, and pragmatically useless at best. Of course, it doesn't follow from any of the aforementioned lines of reasoning that there are no essential properties, but rather just that, if there were, we would seem to have an epistemic barrier in knowing which were essential and which accidential, indicating that the idea was more of a useful human construct than an objective reality (as William James said in his Principles of Psychology, while I am writing on a piece of paper it is essential that I concieve it as a "surface for writing", but if my fire is getting low it is essential that I concieve it as "fuel for a fire"). My views are however being reconsidered for some of the proceeding reasons.

The first thing I began to think about was the fact that I liked to say "I don't believe in essential properties in terms of an abstract essence, but I do believe in 'essential properties' in the sense that there are probability distributions among species". What I meant was that, while it was incoherent to view essential properties in their real, metaphysical sense, it nonetheless was obviously the case that different species tended to actualize certain properties taken as a whole, and that these properties could be conveniently classified as "essential". I thought I could leave it at that, but I was wrong. The reason is that the fact that species, as a whole, tend to manifest certain unique behavioral and biological properties is itself in need of explanation. It would be silly to take facts about the near universal actualization of particular properties by certain organisms to be a trivial fact that had no source in some deeper reality. When one asks the question, "Why does species x tend to manifest certain regularities; tend to regularly behave, reproduce, and interact with the world in certain ways?", the simplest and most obvious answer is because it is in their nature to do so. Sure, not every member of a species fully actualizes their nature, but they are nonetheless part of a wider population, and ancestry, that almost uniformly does manifest such a nature, and thuse can be reasonably said to belong to that species, though in a disfunctional way.

To further clarify this point about probability distributions, we should look at an everyday example of a probability explanandum and its explanans. Take the example of coin-flipping. Say you're flipping a coin that repeatedly comes up heads (say the ratio of heads-to-tails is 20:1). The fact that heads comes up 20:1, an odd and repeated tendency in favor of heads, cries out for explanation. We reasonably infer that the coin has been rigged to land on heads. Put another way, we have infered that despite the fact that it sometimes lands on tails, it has been made to be of such a nature as to land on heads. And most probability-based inferences share this general structure: That the probability distribution D implies something about the nature of the object giving rise to D. Put still another way, there is always an explanans logically prior to facts about probability distributions.

Once I realized that such probability distributions were best explained by something like a nature, I began to think about the epistemic problem again: How do we decide what specifically belongs to that nature and what doesn't? Of course, we could never know all that a thing's nature enompassed (because, as James said, we are a "wellspring" of properties), but it seems to be that there are what you might call "essence-indicators" in things that point us in the right direction.

Inextricable from the idea of a nature is the idea of behaving, or more broadly functioning in a certain way. To say that a thing has a nature, or an essence, is to say that it has parts which function in a unique way to produce certain normative behaviors and properties. So, then, if there is a part of that functional system which only makes sense as a part of the integrated whole, then we may reasonably say that it constitues an aspect of the thing's nature. To work this idea out, let's take the example of an organism having the properties of both (a) having a birthmark, and (b) having a leg (let's say it's a human being for argument's sake). If you didn't know anything about human beings, and you happened upon a newborn child, you would observe its many characteristics, including (a) and (b). Upon examining (a), it would not be apparent to you that it served any unique or functional purpose as part of the organismal whole. (a) is merely some sort of mark, with no apparent relation to any other part of the body to make sense of it. (b) on the other hand can be seen to function cooperatively with many other parts of the body such that the other parts make sense of what it is. So, for example, you see that the infant has not only this one leg, but two! And that these both appear to work together as the infant first crawls, and then walks in the world. The difference between (a) and (b) is further highlighted if you consider a case in which the child was born with only one leg. If the child were born with only one leg, then it would appear obvious, even to those not familiar with human beings, that the remaining leg would make more sense if the other were present; make more sense in such a way as to define each individual leg as "a thing that walks", although this property is only manifest when functioning in cooperation with another leg. These privations I will call "functional traces".

The presence of functional traces, then, serve as essence-indicators. To take another example, think of a child being born with only one eye, and where the other eye should have been there is an eye-socket with nerves hanging from the visual cortex down in it. The fact that you can see the other eye filling the socket with nerves hooked into the back of it is a functional trace relative to seeing how the other eye properly functions.

So now we have a general picture emerging that makes the existence of discrete natures appear more plausible. The ontological reality of natures is suggested by and in the regularities of the behaviors and properties that species exhibit. The epistemic problem of how we may reliably demarcate the essential from the accidential properties is resolved by the knowability, through empirical investigation, of functions and functional traces. I think, however, that we can make this account clearer still by utilizing a set-theoretic framework.

I want to propose that we think of the essence (or nature) of a thing as a set of properties. Call this set S. Additionally, call any species member K, so that S will be a set had by K. S will be formally defined as the set that contains all and only those properties which (a) are normatively actualized by the species to which K belongs, (b) are, in principle, empirically discoverable as functional traces in a state of disfunction, and (c) are a set of properties such that, if some member of another species K* possessed S as a proper subset, K* would share a nature with K (this last property being S's self-membership). Lastly, I should clarify that a species possessing a property-set that overlaps with S is not sufficient for a nature to be shared, but the species must possess S as a proper subset. I may have more to say about this in future entries, but it suffices for now to a give a rough idea of how essences may be formally defined.
Thus it is that, while not fully endorsing a view of natures, I now view it as a more formidable philosophical view than I had before. But, of course, there are many other issues which need to be addressed in deciding whether things have natures; issues concerning the coherence of a bare substratum, teleonomy vs. teleology, and the like. I may explore such problems in future writings, but I hope the present outline has been initially enlightening.