Mackie's Original Objections
Mackie's Criticism of Swinburne
In Chapter 8 of The Miracle of Theism, Oxford philosopher J. L. Mackie summarizes the objections raised by Hume to the design argument in Dialogues Concerning Natural Religion. Mackie finds five objections there:
I also argued in lectures 13 and 14 that objection number three depends on an anthropomorphic conception of God, which in turn depends on a similarity-based version of the design argument. The non-representational, non-discursive model of the divine mind sketched in these lectures seems to rebut the force of Hume's objection.
We will take up Hume's fourth point in lectures 19 through 24, when discussing the problem of evil. We can observe that this objection has nothing to do with the question of whether an infinite creator exists, but is only relevant to the question of whether we can attribute moral goodness to this creator.
The fifth point listed by Mackie is one that must, I think, be conceded. We cannot expect to get all that we need from a religious point of view from the design argument alone. Nonetheless, the design argument can serve to "prepare the understanding for theological knowledge", as Kant puts it. Understanding that there is an infinite, intelligent and purposeful creator makes theological explanations of other phenomena, such as religious experience or reports of miracles, more plausible than they would otherwise be.
Mackie raises three additional objections to the design argument. First, he argues that the argument from the anthropic coincidences overlooks the possibility of exotic life, life very different from that with which we are familiar. We have already discussed this objection in lectures 11 and 12.
Second, Mackie argues that we do not find a regular association between the property of being ordered and adapted to certain ends and the property of having been constructed by human designers. Man-made machines are far less prevalent than naturally-occurring organisms, which, according to the Darwinian theory of evolution, arose without the agency of an intelligent designer. This objection depends on miscasting the argument from design as an argument based on observing a correlation between order and design, but the teleological argument from design is based on no such observed correlation. As I argued in the parable of the Stranger, one could infer the existence of design from the anthropic coincidences, even if one had no other experience with design and its effects. Moreover, I have argued that, on the Darwinian story, nature is invested with a kind of intelligent agency.
In Mackie's third objection, he argues that theism is not a good scientific theory, nor is it a promising scientific research program. A good scientific theory generates novel predictions, or at least, as in the case of retrospective theories like Darwinism, generates "differential explanations" of otherwise inexplicable details. Mackie argues that theism does neither, because the theistic hypothesis is so vague, and because God is, by his very constitution, unpredictable and uncontrollable.
In addition, Mackie argues, theism is not a promising research program. A research program is not a scientific theory, but a kind of template or recipe for generating theories. Darwinism might best be thought of as a research program, with specific theories about the ancestries of various species or the environmental pressures shaping various adaptations as the kind of specific theories that the program generates. Mackie argues that theism generates no such interesting theories.
There are three replies to be made to Mackie's final objection. The first is simply, so what? The theist can readily admit that theism is neither a scientific theory nor a scientific research program, but still insist that theism is the most reasonable conclusion to draw upon the available evidence. For Mackie's point to count as a serious objection, Mackie would have to argue that all our knowledge is scientific knowledge, that apart from science we have no knowledge whatsoever. The claim by theists like Swinburne and myself is only that the case for theism is analogous to the case for a well-confirmed scientific hypothesis, not that the two are exactly the same thing.
Second, it could be argued that theism does offer differential explanations of just the kind Mackie requires. For example, theism offers an explanation of the anthropic coincidences, which would otherwise be inexplicable. In addition, there are a great number of specific facts about the world for which theism offers explanation, including a variety of religious experiences and remarkable historical events, such as the survival of the Jews and the extraordinary rise of Christianity in the first century. Theism offers explanations for the existence of the moral sense, especially the absolute and authoritative character of its intimations. Theism explains the sense we have of being more than just matter in motion, and our need for transcendent hope and meaning. Finally, Michael Behe, William Dembski and others have argued that theism offers an explanation for the existence of irreducibly complex biological mechanisms, mechanisms whose function is dependent on the cooperation of so many parts that it is implausible to suppose that the mechanism could have arisen from chance, even when the possibility of coopting parts of pre-existing mechanisms is taken into account.
Third, one could argue that the development of theology demonstrates that theism is a progressive research program. John Henry Newman's masterful work, The Development of Doctrine, documents the progressive character of Christian theology, and similar claims could be made for many other traditions.
At the end of Chapter 8, Mackie criticizes an early version of Swinburne's design argument, one appearing in Swinburne's book, The Existence of God. This version of the argument made no mention of the anthropic coincidences, and instead depended entirely on the simplicity, elegance and universality of the world's natural laws, which Swinburne called the world's "temporal order" (i.e., its orderliness across time). Swinburne claims that, in the absence of the existence of a creator with a marked preference for order and simplicity, it would be antecedently very surprising to find the world as simple and orderly as it in fact is. Mackie demands a basis for the claim that the prior probability of such regularity is low.
Swinburne would respond that there are many more ways for the world to be chaotic than there are for it to be simple and regular. Simple and regular worlds seems to occupy a small region of logical space. In addition, this region can be specified simply and without reference to how things actually are. Consequently, this simplicity is in need of some explanation.
Mackie attempts to show that Swinburne's argument is internally inconsistent. As we have, the process of reaching theoretical conclusions is governed, according to Swinburne, by a very pronounced bias on our part toward simplicity. When we find one and only one very simple explanation of a body of unlikely data, we are drawn irresistibly toward the simple explanation as the correct one. Mackie argues that this bias toward simplicity means that we start out by assuming that, in all likelihood, the universe really is very simple. This means that the prior or antecedent probability that the world would be simple is actually quite high, not low, as Swinburne's argument from "temporal order" requires.
Thus, Mackie charges Swinburne with trying to have it both ways. In arguing that the simplicity of the world is in need of explanation, Swinburne is assuming that the prior probability of this simplicity is quite low. In arguing that we should accept theism because it is the simplest explanation of this temporal order, Swinburneis assuming that the prior probability of simplicity is high.
This is a rare case in which Mackie has simply made a logical or mathematical blunder. There is no inconsistency in Swinburne's position. Let "S" represent the proposition that the temporal order of the world is quite simple (at least as simple as that of the actual world). What Swinburne's account of scientific inference requires is one of the following principles:
(*) Whenever two hypothesis H1 and H2 are of comparable scope and precision, and H1 is much simpler than H2, then H1 is more probable than H2.
(**) If H is a very simple hypothesis, then there exists a possible body of data E such that P(H/E) > 1/2.
(***) If H is a very simple hypothesis, then for every number r greater than zero, then there is a body of possible data E such that P(H/E) > 1 - r.
The third principle, (***), is the strongest. If we assume it, and we assume that the structure of the world really is quite simple, then it follows that we can come to know what the structure of the world is with a degree of confidence that approaches absolute certainty. The other two principles would give us somewhat lesser degrees of certainty. However, even principle (***) does not entail what Mackie thought must follow, namely, that the prior probability of S is high.
What principles (*) through (***) guarantee is that on any point by point comparison, the simpler hypothesis must be more probable than the more complex hypothesis. However, it does not follow that the sum of the probabilities of all the simple hypotheses (which is equal to the probability of S) must be greater than the sum of the probabilities of all the complex hypotheses (which is equal to the probability of not-S). To think that this does follow, as Mackie did, is to commit the fallacy of composition (assuming that what is true of each of the parts must be true of the whole). In a one-to-one comparison, the simpler hypotheses are more likely than the complex ones. However, there are far more complex hypotheses than there are simple ones. Hence, the sum of the probabilities of all the complex hypotheses could be much greater than the sum of the probabilities of the simple ones.
This same point can be made in a different way. In addition to talking about S, the hypothesis that the structure of the world is in fact quite simple, we could also talk about the proposition E-S, the proposition that the body of observed data is as large as the body of actually observed data and is consistent with S. Given the bias toward simplicity that is built into Swinburne's account of scientific inference, we can conclude that the conditional probability of S given E-S is quite high. However, it does not follow that the probabilityof E-S is itself at all high -- the prior probability of E-S is presumably very low. Consequently, even though the conditional probability of S on E-S is high, the prior probability of S itself might be very low, as Swinburne's design argument requires.