A. The Arguments Formulated
In chapter 13 of Atheism: A Philosophical Justification, Michael Martin argues that Salmon’s arguments can be expanded to provide a powerful case for atheism:
If Salmon’s arguments concerning the unique properties of God are restated and expanded, they provide a powerful inductive case for positive atheism in the narrow sense. The theistic God is an all-powerful, all-knowing, all-good, disembodied person who created the universe out of nothing. If it can be shown that, in the light of the evidence, such a being is improbable, then disbelief that the theistic God exists is justified. Consequently, positive atheism in the narrow sense is justified.
(1) In terms of our experience, created entities of kind K that have been examined are always (or almost always, or usually) created by a being (or beings) with property P.
(2) The universe is a created entity.
(2a) If the universe is a created entity, it is of kind K.
(3) The universe was created by a being with property P.
(4) If the theistic God exists, then the universe was not created by a being with property P.
(5) Therefore, the theistic God does not exist.
Martin then presents five specific teleological arguments for atheism, all of which adopt the general form. In the interest of brevity, for each argument, I will list the argument’s name, the kind K, and the property P.
|Argument Name||Kind K||Property P|
|The Argument from Embodiedness||all created entities||bodies|
|The Argument from Multiple Creators||all large and complex created entities that we have so far examined||a group of beings working together|
|The Argument from Apparent Fallibility||entities with seeming errors or mistakes||fallible being|
|The Argument from Apparent Finiteness||all created entities||being with finite power|
|The Argument from Preexisting Material||all created entities||being(s) using preexisting material|
B. Assessment of Martin’s Arguments
The inference from (1), (2), and (2a) to (3) in the general form is an argument from positive analogy; it moves from a generalization about an observed property (O) in a sample to an inferred property (I) in a specific instance. As William Gustason explains, “The most important criterion in judging the strength of an analogical inference is the relevance of the properties in the positive analogy to the inferred property.” O is “positively relevant” to I if the presence of O increases the probability that I is also present. This suggests a simple test for evaluating Martin’s teleological arguments: for each argument, we can ask, “Does the universe’s having observed property O increase the probability that the universe also has inferred property I?” If the answer is “yes,” then the analogical argument succeeds; if “no,” then the argument fails.
Let us consider each of Martin’s teleological arguments one-by-one.
1. The Argument from Embodiedness
Observed property (O): created
Inferred property (I): created by a creator with a physical body
We know a lot more about created entities than the fact that they are created entities. We also know their relationship to spacetime. Created entities in the universe are created in space and time; known entities created in space and time are created by embodied creator(s). In contrast, according to current scientific knowledge, the physical universe, including spacetime itself, began with the big bang. How does the hypothesis of an embodied creator fit with the creation of spacetime itself? Either there is a multiverse or there isn’t. If there is no multiverse, then an embodied creator seems to be a physical impossibility, since an embodied creator’s physical body would need to be part of the physical universe and hence could not exist ‘outside’ of the universe to create it. If there is a multiverse, which is far from certain, then it is at least possible there could be embodied persons in other universes. But even if we assume there is a multiverse, there is nothing in our current scientific knowledge which makes it probable that an embodied person in one universe can create another physical universe. In other words, the universe’s having O does not increase the probability that the universe also has I. In conclude, then, that the argument from embodiedness fails.
2. The Argument from Preexisting Material
I: created from preexisting material
As Martin notes, this is not an argument against theism, but against any version of theism which affirms creation ex nihilo, such as Christian theism.
While this argument has an intuitive appeal to me, I find it difficult to have much confidence in this argument, for reasons similar to what I wrote about the argument from disembodiedness, namely, it’s unclear how the concept of “preexisting material” is supposed to fit with the creation of the universe itself. (Where was that “preexisting material” before the universe was created?) Indeed, if I were to formulate an argument from positive analogy with preexisting material, I would be inclined to reach a different conclusion: “The universe was not created.” But the argument from preexisting material presupposes that the universe was created. The act of “creating a physical universe by a disembodied being” is so unlike anything in our experience it is difficult to have confidence in an argument of this sort.
3. The Argument from Multiple Creators
O: large and complex
I: created by multiple creators
It’s far from clear how to assess whether the universe’s having O increases the probability that the universe also has I, i.e., how the universe’s being created provides evidence for polytheism and against monotheism. On the one hand, known instances of the creation of large and complex entities in the universe do require multiple creators. On the other hand, those creators are embodied beings (i.e., humans) creating entities in the universe. It seems to me that the creation of the universe requires unimaginably more power than the creation of entities in the universe. Again, the act of “creating a physical universe by a disembodied being” is so unlike anything in our experience it is difficult to even estimate the number of disembodied beings needed to create a universe.
For example, let a “jupiter”–note the lower case ‘j’–be a unit of measure for the power or abilities of a being. The Roman god Jupiter (and his Greek counterpart Zeus) have 1 jupiter of power, Jedi Master Yoda has 10-4 jupiters, and ordinary adult human beings have only 10-6 jupiters. In contrast, we may suppose an omnipotent being, viz., the God of classical theism has infinite jupiters. On the assumption that the creation of the universe by a disembodied being is intelligible, how many jupiters are needed to perform such an act? 1? 1040? An infinite number? (Suppose we knew that it were metaphysically necessary that only beings with infinite jupiters can create physical universes. In that case, we would have a strong analogy to a single, omnipotent creator, not multiple creators.) The upshot is this. Since we do not know the ‘average jupiters’ for an unspecified disembodied mind and we don’t know the ‘minimum jupiters required to create a universe,’ I am unable to see how we could ever have confidence in an analogical argument of this sort.
4. The Argument from Apparent Finiteness
I: created by a being with finite power
I have the same worry about this argument as I do about the argument from multiple creators. Again, since we do not know the ‘average jupiters’ for an unspecified disembodied mind and we don’t know the ‘minimum jupiters required to create a universe,’ I am unable to see how we could ever have confidence in an analogical argument of this sort.
5. The Argument from Apparent Fallibility
O: created entities with seeming errors or mistakes
I: created by a fallible being
This argument reminds me of William Rowe’s first version of the argument from evil, which crucially depends upon an inference from ‘apparently pointless evils’ to ‘actually pointless evils.' Similarly, the argument from apparent fallibility seems to depend upon an inference from ‘created entities with seeming errors or mistakes’ to ‘created entities with actual errors or mistakes.’ Because these arguments are so similar, it seems to me both arguments stand or fall together. Either one is persuaded by both Rowe’s argument from evil and Martin’s argument from apparent fallibility or one rejects both arguments.
Arguably the strongest objection to Rowe’s argument is that it depends upon a so-called “noseeum inference”: the inference from “I can’t see X” to “there is no X.” Rowe makes such an inference when he argues that our inability to see the point for some suffering makes it reasonable to believe that there is no point for that suffering. Rowe’s critics, most notably Stephen Wykstra, have argued that such an inference is invalid: given the limited cognitive abilities of humans, we should not expect to see the point of some suffering even if there is a point. Similarly, the argument from apparent fallibility also seems to implicitly depend upon a noseeum inference: the inference from ” I can’t see the point for some apparent error or mistake in the universe’s design” to “there is no point for some apparent error or mistake in the universe’s design.”
 Michael Martin, Atheism: A Philosophical Justification (Philadelphia: Temple University Press, 1990), 321.
 Martin 1990, 322-24.
 Martin 1990, 324-26.
 Martin 1990, 326-28.
 Martin 1990, 328-29.
 Martin 1990, 330.
 William Gustason, Reasoning from Evidence: Inductive Logic (New York: Macmillan, 1994), 61.
 Martin 1990, 330.
 William L. Rowe, “The Problem of Evil and Some Varieties of Atheism,” American Philosophical Quarterly 16 (1979), 335-41; reprinted in The Evidential Argument from Evil (ed. Daniel Howard-Snyder, Indianapolis: Indiana University Press, 1996), 1-11.
 Stephen John Wykstra, “Rowe’s Noseeum Arguments,” The Evidential Argument from Evil (ed. Daniel Howard-Snyder, Indianapolis: Indiana University Press, 1996), 126-50.