Monday 7 December 2009

Towards a Teleological Logic (Part 4)

In this post, I wish to consider a pair of objections to the account of a T-world defended thus far. One putative difficulty with characterising T-worlds as possible worlds in which every telos is realised is that it seems to preclude compensatory and/or conflicting purposes. The notion of a compensatory purpose applies to teleological objects that have the telos of “filling in” for when some other teleological object fails to realise its telos. For example, we can imagine a system equipped with an emergency self-destruct sequence that only initiates if there is a failure in all other safety protocols. If we conceive of T-worlds as worlds in which every telos is realised, then such a self-destruct mechanism will never have the opportunity to realise its telos since there will never be the required failure elsewhere in the system. This suggests that in a given T-world, compensatory purposes are never realised. However, if compensatory purposes remain unrealised, then a T-world cannot really be a world in which every telos is realised. Thus, the possibility of compensatory purposes appears to threaten the concept of a T-world with incoherence.

One possible reply to the above objection, which will ultimately prove inadequate, is to distinguish between cases in which some X does not have the opportunity to fulfil the function for which it is designed and cases in which X is presented with such an opportunity, but fails to do so. Moreover, we may say that a teleological object only fails to realise its telos in the latter case. This suggestion appears to have some intuitive traction since few would regard a safety mechanism as somehow defective simply because it never had the opportunity to perform its function. On this view, some world Γ counts as a non-T-world just in case some X in Γ is presented with an opportunity to realise its telos and yet fails to do so. Thus, far from implying the failure of compensatory purposes, T-worlds virtually guarantee that such purposes are realised by removing the antecedent conditions for their failure. Consequently, the concept of a T-world remains coherent, even if there are teleological objects with compensatory purposes.

The above reply seems right, as far as it goes. But there are at least two reasons for thinking that it does not go very far. First, while it appears to avoid the problem posed by compensatory purposes in the case of artefacts, it is not clear that the reply generalises to biological systems. We can certainly imagine a biological system or process that has the function of “filling in” for some other biological system or process, should the latter fail to realise its telos. Moreover, we can imagine such a system evolving in the actual world since individuals within a biological population with such a system acting as “back-up” would display greater reproductive success than their conspecifics that lacked such “safety nets”. However, it is not clear that such a “back-up” system could ever evolve in a T-world. Since the original system will always realise its telos, the presence of a “back-up” system will never offer any evolutionary advantage to its possessor, and will therefore never become subject to selective pressures. The upshot is that there could never be a biological system with a compensatory purpose in a T-world.

Second, the definition of T-worlds as possible worlds in which every telos is realised seems to be at odds with the fact that there may be conflicting purposes. The notion of a conflicting purpose applies to teleological objects whose telos involves preventing some other teleological object from realising its telos. For example, we can imagine a type of antibiotic whose purpose it is to prevent the DNA found in a particular bacteria from performing its replicatory function. Insofar as we define a T-world as one in which every telos is realised, then both the bacterial DNA and the antibiotic cannot coexist in the same T-world. But since we have assumed a constant domain semantics, and since there are possible worlds in which both the bacterial DNA and the antibiotic exist (e.g., the actual world), this appears to throw the notion of a T-world into jeopardy.

I believe we may overcome the challenge posed by both compensatory and conflicting purposes via a multimodal teleological logic, in which the accessibility relation Ri is indexed to a particular biological system or human artefact, i. Instead of a single “common” accessibility relation R, there is a series Ri, Rj, Rk . . . , indexed to sets of teleological objects (e.g., the set of human eyes, the set of human ears, the set of hammers). The Kripke frame for the corresponding language, L, in which {□i|i ∈ I} represents the set of necessity operators of L, consists of a non-empty set of possible worlds G, and the binary relation Ri, for each i ∈ I. The satisfaction relation for □i is defined as follows:
(5.1) w ⊩ □i φ if and only if ∀u(Ri(w,u) → u ⊩ φ).
According to (5.1), φ is true in all T-worlds relative to some biological system or artefact i just in case φ holds in any world that stands in the relation R to some other world. Instead of speaking of T-worlds in which every telos is realised, we now speak of T-worlds indexed to some biological system or artefact, i, such that i always realises the telos for which it was selected or designed in the actual world.

On this view, the possible worlds in which the antibiotic performs its function constitutes the set of T-worlds indexed to that antibiotic, while the possible worlds in which the DNA of a particular bacterium realises its telos represents the set of T-worlds indexed to that bacterial DNA. Since each teleological object is now indexed to its own set of T-worlds, there can be no conflicting purposes within T-worlds. Thus, the coherence of the concept of a T-world is preserved. The multimodal account also avoids the problem posed by biological systems with compensatory purposes. Since the set of T-worlds indexed to some biological system i includes worlds in which some other biological system j fails to realise its telos, this allows i to increase the reproductive fitness of its host when i serves as a “back-up” system in the eventuality of j failing to realise its telos. The upshot is that on the multimodal account, there is no difficulty posed by cases of compensatory or conflicting purposes.