Reply to Michael Martin on the Cosmological Argument: Part Two

*All citations, unless otherwise noted, are taken from Martin's book, Atheism: A Philosophical Justification.  Martin's words will be in blue.

Martin skips the argument from motion in his treatment of the cosmological argument, which is a bit puzzling, considering how many of the God-attributes are predicated on there existing an Unmoved Mover.  I'll return to this and to bridging the gap between first cause and God in Part Four.  In this section, however, I'm going to respond to Martin's critique of Thomas Aquinas's "second way."

In this argument Aquinas attempts to show that there could not be an infinite series of efficient causes and consequently there must be a first cause. (p. 98)

It's important to understand what is meant by "efficient cause," since this is what distinguishes the second way from the argument from motion (the first way).  In Aristotelian terms, an efficient cause is something that produces an effect.  There are four types of causes: efficient, material, formal, and final.  Take, for instance, the example of a painting.  The efficient cause is the painter; the material cause is the paint; the formal cause is the idea or blueprint of the painting in the painter's mind; and the final cause is the end or goal of the painter, e.g. to produce a painting that exemplifies beauty.

Although this notion of efficient cause is perhaps closer to our modern view of causality than the other Aristotelian concepts of cause he used, there are some important differences.  An efficient cause of something, for Aristotle and Aquinas, is not a prior event but a substantial agent that bring about change. (p. 98)

I'm not sure why Martin thinks that this is different than the modern usage of "cause."  A substantial agent that causes change is an efficient cause whether time is involved or not.  Although, it should be noted that time is a measurement of change.

The paradigm cases of causation for an Aristotelian are heating and wetting.  For example, if A heats B, then A produces heat in B; if A wets B, then A produces wetness in B.  In general, if A [x's] B, then A produces [x]ness in B.  The priority of a cause need not be temporal; a cause is prior to its effects in the sense that the cause can exist without the effect but not conversely. (p. 98)

This is correct.  Supposing that a house has existed for all eternity, and has an infinite past, the parts of the house remain standing only because of its foundation.  The foundation may exist without the house's standing parts, but not the other way around.  If at any point the foundation is removed, then the house will collapse.

It is important to realize that Aquinas's argument purports to establish a first cause that maintains the universe here and now.  His second way is not concerned with establishing a first cause of the universe in the distant past. (p. 98)

Again, this is correct.  According to Thomas, the universe requires a first cause with respect to sustaining causality, regardless of whether it has a first cause with respect to originating causality.

Indeed, he believed that one could not demonstrate by philosophical argument that the universe had a beginning in time, although he believed that it did.  This belief was a matter of faith, something that was part of Christian dogma, not something that one could certify by reason.  Thus he was not opposed on philosophical grounds to the universe's having no temporal beginning.  As the above quotation makes clear, he believed that the here-and-now maintenance of the universe could not be understood in terms of an infinite causal series. (p. 98)

We will take a look at the so-called "kalam cosmological argument" later, which purports to show that the universe must have had a beginning.  While Thomas did not believe this could be demonstrated on philosophical grounds, his contemporary, Bonaventure, believed otherwise.

Two analogies can perhaps make the distinction between temporal and nontemporal causal sequences clear.  Consider a series of falling dominos [sic].  It is analogous to temporal causal sequence.  Aquinas does not deny on philosophical grounds that infinite sequences of this sort can exist.  But now consider a chain in which one link supports the next.  There is no temporal sequence here.  The sort of causal sequence that Aquinas says cannot go on forever but must end in a first cause is analogous to this. (pp. 98-99)

The latter is similar to the example I gave above of a house's ability to stand.  The question is whether the house's foundation could be grounded in an infinite regress of sustaining causes, or whether there must be a first cause.

The same problems that plagued the simple version of the argument plague this more sophisticated version.  The first cause, even if established, need not be God . . . (p. 99)

As mentioned previously, we will take a further look at this in Part Four.

[A]nd Aquinas gives no non-question begging reason why there could not be a nontemporal infinite regress of causes. (p. 99)

Although Martin makes this claim, he never actually cites any of the arguments Thomas presents against such a regress of causes.  One would think he would state clearly what the arguments are, and then explain why they are question-begging.  Martin adds a footnote:

Kenny argues that Aquinas's views on nontemporal causal sequences are closely related to theories of medieval astrology and that his argument that an infinite nontemporal causal series is impossible rests on an equivocation between "first=earlier" and "first=unpreceded." (p. 493)

Martin doesn't elaborate on either of these points, but as we will see shortly, Thomas's arguments require neither astrology nor an equivocation of terms.

However, Rowe suggests that Aquinas's views do not rest on medieval astrology but on a metaphysical analysis of existence and causation. (p. 493)

Let's take a moment to let this sink in.  Although Martin only mentions this in one sentence, this is an extremely crucial point.  Grounding Thomas's argument is an Aristotelian distinction between being (existence) and essence (a thing's nature).  If there is no first cause (being itself subsisting), then why does anything at all exist?  Presumably the difference between a unicorn that's real and a unicorn that's not real is that the former has being, regardless of their similarity in essence.  If being itself does not exist, then there is no difference between the two.  After all, to differ by non-being is to differ by nothing, and to differ by nothing is to be identical.  Since there is obviously a difference between a reality and a non-reality, it follows that being exists.  Since being itself is independent of any further being, it follows that being is the first cause of all other existing essences.  I explain this further here.  This is true even supposing there is an infinite regress of sustaining causes, for the regress itself is dependent upon being.

Martin never addresses this argument, which is found in Thomas's work, De Ente et Essentia ("On Being and Essence").

Nevertheless, Rowe argues that Aquinas's actual argument is question-begging and tries to reformulate the argument in a way that is not.  Rowe's reformulation presupposes the principle of sufficient reason. . . . As Rowe argues elsewhere, we have no reason to suppose that the principle of sufficient reason is true or that we can assume that it is true. (p. 493)

Rowe's view is a bit more nuanced than Martin leads on, in my estimation.  However, an appeal to the PSR is unnecessary.  We'll return to the PSR as we examine Martin's critique of the Leibnizian cosmological argument.

This latter is an especially acute problem.  Unless some relevant difference is shown between a temporal and a nontemporal infinite series, Aquinas's claim that an infinite temporal sequence cannot be shown to be impossible by philosophical argument seems indirectly to cast doubt on his claim that philosophical argument can show the impossibility of a nontemporal causal series. (p. 99)

Martin stops his treatment of the second way here, without ever explaining what Thomas's reasons are or might be.  In the Summa Contra Gentiles, Part One, Chapter 13, Thomas provides arguments that illustrate a relevant difference between a temporal versus a nontemporal series of causes.  More precisely (since change presupposes time), Thomas provides an argument against an infinite regress of sustaining causes.

One reason is that even supposing that the past is infinite, it is still composed of finite intervals.  This is true even granting that time and space are continuous.  After all, there is a real difference between 4:00 pm and 5:00 pm, and there is a real difference between the locations of New York and Chicago.  Now, for each finite interval of time, the regress of sustaining causes begins anew.  Since it is impossible to form an actual infinite by successive addition whenever one begins counting, it is impossible for the regress of sustaining causes to be infinite.  Therefore, even if the past is infinite, the regress of sustaining causes during each finite interval of the infinite past cannot likewise be infinite.

Martin himself admits this in his critique of the kalam cosmological argument:

"[A]n actual infinity can be constructed by successive addition if the successive addition is beginningless." (p. 105)

Somewhat poetically, Martin unwittingly provides support for the notion that there cannot be an infinite regress of sustaining causes.  This is due to the nature of sustaining causes having a beginning at each finite interval.

In sum, Thomas's second way looks something like this:

1. Every dependent thing has a sustaining cause. (Premise)

2. Either an independent first cause exists, or else there is an infinite regress of sustaining causes. (Implied by 1)

3. There cannot be an infinite regress of sustaining causes. (Premise)

4. Therefore, an independent first cause exists. (From 2 and 3)

These aren't the exact terms Thomas uses, but they suffice for our purposes.  The argument is also immune to the "what causes the first cause?" objection, that Martin thankfully does not appeal to.  The causal premise is restricted to dependent things, and is not descriptive of everything necessarily.