My previous post on evaluating traditional theistic arguments has generated a lot of discussion! Thanks to Jonathan, Lewis, and Clayton for helping to clarify some issues with my initial presentation.
Most of the discussion centered on the teleological argument. I'm not sure if that's just because I presented it first, or because it was the most problematic. (Lewis, who I hope doesn't mind me quoting him, told me offline yesterday that the teleological argument is the 'nutsiest' of all theistic arguments because it requires the theist to take the evidential argument from evil seriously and makes it difficult to use skeptical theism as a response to it. I agree that it has these problems, but I think classical theists are faced with these problems to begin with. That's a topic for another day.) Anyway, let me try to make my version of the argument a little more precise, and consider some objections. (The most important objection, I take it, is that we see the world through teleology-colored glasses, as it were; more on that below.) In trying to make the argument the more precise, I will necessarily rely on more controversial philosophical premises than my quick sketch did, but that's always the way it goes with precise arguments.
I begin with some terminology. Philosophers distinguish between perceiving an object and perceiving that a proposition is true. Perceiving that p is often (though not always) held to be sufficient for knowing that p. Furthermore, perceiving that p is factive - that is, it is part of the meaning of 'perceives that' that if S perceives that p, then p. What is not clear to me is how exactly the belief that p is supposed to be formed, and how this is related to the agent's perceptual apparatus. Some philosophers think the perceptual apparatus actually forms the belief, but I have imbibed too much classical empiricism to find this plausible. So let me (controversially) define perceiving that as follows:
S perceives that p =df. S has a veridical perceptual experience and, by the normal operation of S's cognitive apparatus S forms the true belief that p on the basis of that perceptual experience without conscious inference.
It seems to S that p =df. S has a perceptual experience and, by the normal operation of S's cognitive apparatus S forms the belief that p on the basis of that perceptual experience without conscious inference.
S is reasonable in performing cognitive act φ =df. In φing, S approximates the ideal of rationality as nearly as can be expected given S's limitations.
Reasonableness of Seemings (RS): If (i) it seems to S that p and (ii) S has no evidence that not-p and (iii) S has no evidence that this particular seeming is untrustworthy, then S is reasonable in retaining p.
Now, I take it that the reason (RS) is true is that a seeming confers some sort of positive epistemic status on a proposition. This status may not be especially strong. As mentioned above, many philosophers do think perceiving that is sufficient for knowledge, but I think that might be because they have a different conception of perceiving that than I do. Or perhaps seemings only confer full justification when they are seemings of certain sorts - e.g. when an object seems to have a particular perceptible quality. So I think the epistemic status here is weaker than full justification. However, it cannot be too weak for, under certain circumstances, it is sufficient for the reasonableness of belief. For comparison, suppose I am investigating a murder, and I learn that Smith was not at home at the time of the murder. This confers some sort of positive epistemic status on the proposition that Smith committed the murder, but surely I would be unreasonable to positively believe that Smith committed the murder if I had no further evidence. So seemings confer stronger positive epistemic status than circumstantial evidence.
Now, here is the next problem. Lewis has shown that the argument schema for the transmission of epistemic status in the previous post won't work. What needs to be said instead is that when an agent comes to know that p entails q (where p may be a long conjunction), some sort of harmonization of the agent's confidence in those propositions ought (rationally) to occur. Bayesianism provides a formal framework for this kind of adjustment, but non-Bayesians also agree that it needs to occur. For instance, if p is certain, then knowing that p entails q is sufficient for knowing that q. (Though we have to take into account one's confidence in the entailment claim as well.) However, and this is the main problem with my previous formulation, sometimes the adjustment ought to go backward. This is what happens in a reductio ad absurdum: q is absurd, so the fact that p entails q ought to undermine our belief in p. However, in general, being entailed by premises which have positive epistemic status will improve the epistemic status of a proposition. So here is my new principle:
Transmission of Reasonableness (TR): If (i) S believes that p and (ii) S knows that p entails q and (iii) S is reasonable in continuing to believe that p after learning that p entails q, then S is reasonable in believing that q.
Now, here is the actual argument:
It seems to many people (in my technical sense of 'seems to') that (1) is true, and (2) is fairly plausible. So (3) will be reasonable unless either (a) condition (ii) or condition (iii) of (RS) is not satisfied, or (b) the independent evidence against (3) is strong enough that the argument should be taken as undermining the premises rather than establishing the conclusion. Now, if the theistic hypothesis is true, then (3) is true. So, strong evidence against (3) would be strong evidence against theism. However, evaluating the plausibility of theism in light of our total evidence has proven to be a cognitively demanding task. So it seems likely that if the argument fails for reason (b), there is still some real person somewhere who is reasonable, in my sense, in increasing her confidence in theism as a result of this argument. Accordingly, I won't spend any more time on (b).
(a) is, I take it, the more interesting case. The argument will be undermined due to condition (a) if either (ii) the agent has evidence that, despite appearances, these objects and events do not have purposes, or (iii) that teleological seemings are unreliable. Of course, I don't claim that if one of these conditions is satisfied the agent would necessarily be unreasonable in accepting (1), or even in accepting (1) on the basis of the seeming, but only that if either (ii) or (iii), then the agent would be unreasonable to accept (1) on the basis of the seeming alone.
In the comments on the previous post, Jonathan gave reasons for supposing that teleological seemings fail both condition (ii) and condition (iii). On condition (ii) he said:
There are things (such as living creatures) which appear to have purpose but which in fact don't, they are the result of undirected evolutionary processes. Therefore there are things in the universe which have the appearance of purpose (of design) but where that appearance is an illusion.
One way of [undermining the 'watchmaker' analogy] might be to argue that watches have never been observed to reproduce themselves, whereas plants and animals have been observed to do so, and the production even of stars and planets by natural causes has been observed, and in biological reproduction, changes often result so that natural selection and survival of the fittest seem able to produce the illusion of telos; animals seem to have the survival of the species as their purpose even though, in the absence of a designer, there is no actual purpose. This objection is not particularly successful. Imagine, by way of analogy, that watches were seen to reproduce themselves naturally and to improve over time: sundials collected parts from their environment which were used to construct pendulum clocks, which collected parts from their environment which were used to build main-spring clocks, etc., until, over time, we arrived at quartz wristwatches and, ultimately, atomic clocks. Would not this cry for explanation in terms of intelligence even more than a single watch found on its own? Doesn't this make the case for design better rather than worse?
Jonathan supports the claim that condition (iii) fails as follows:
A cognitive illusion commonly suffered by humans is mistakenly to discern agency where in fact there is none, where there is only randomness. This has been demonstrated in innumerable psychological experiments.
However, this line also begs the question. For it presupposes the ability of the psychologists to determine that the teleological seemings experienced by the subjects are false. However, if the theistic hypothesis is correct, then there really is teleology everywhere. So without assuming naturalism, one can only use these experiments to show that human beings see teleology everywhere, not that they incorrectly see teleology everywhere.
There is, however, one more point against the satisfaction of condition (iii) which both Jonathan and Clayton brought up, and which I take to be far more interesting and difficult than these first two lines. That is the question, what would a purposeless universe look like? If there is no possible world that wouldn't seem to humans to be a product of design, then, teleological seemings are probably worthless. I recognize the force and importance of this objection and will perhaps evaluate it in more detail on another occasion. I think, however, that this is enough for today.Posted by Kenny at January 29, 2010 9:35 AM
Return to blog.kennypearce.net