'Five Ways' for Proving God...and Why They Fail.
But fear not, there is hope for modern reformulations.
One of Thomas Aquinas’ celebrated contributions to philosophical theology is his five-part case for the existence of God: The Five Ways. These are five arguments for God’s existence drawn from simple empirical observations. I explain and assess each argument below, followed by two observations about Aquinas’ cumulative case for theism. You won’t want to miss even one sentence of this majestic paper.
Each of the five ways (1) draws attention to an empirical fact – motion, efficient causation, contingency, gradation, and the directedness of the natural world, respectively – and then (2) connects those observations to theism. If successful, these arguments give reason to believe in an unmoved mover, an uncaused first cause, a necessary being that is the source or ground of all other contingent and necessary beings, a being perfect in goodness, truth, and nobility, and a mind that imposes direction upon the natural world. There are two features of Aquinas’ case worth mentioning before diving in. First, these arguments are a posteriori. They draw on at least one empirical observation: things are in motion, efficient causes exist, there are contingent beings, and so on. Of course, Aquinas utilizes plenty of metaphysical principles to get from those observations to theological conclusions - but nevertheless, without those initial empirical claims, none of the Five Ways could get off the ground. Aquinas draws a similar distinction between reasoning from effect to cause (e.g., the grass is wet so it must have rained) and from cause to effect (e.g., the room must be warm because there is a fire inside). The Five Ways are exercises in the former.
Second, the Five Ways are only one part in Aquinas’ broader natural theology. Both expositions and critiques of the Five Ways often treat them as if Aquinas intended them to be completely self-sufficient defenses of theism, but Aquinas himself almost certainly did not think of them in this way. So, for example, Oppy (2006) objects that Aquinas does not prove in any of the Five Ways that there is only one unmoved mover, efficient cause, ground of necessity, and so on; for all we know, there are many. That criticism is misplaced. Aquinas addresses this issue of polytheism versus monotheism in another question in the Summa Theologiae. In evaluating the Five Ways, then, one must keep Aquinas’ modest intentions in mind lest one takes them to prove something they never intended in the first place.
#1 Motion
In the First Way, Aquinas argues that the fact that things move (or ‘are moved’, depending on how one translates the Latin word movere) entails that there is an unmoved mover that is the source of all other motion. By ‘moves’, it is important to realize that Aquinas has more than mere change in spatial location (so-called ‘local’ motion) in mind. While he certainly does intend this obvious kind of motion, he also intends changes in quantity (e.g., gaining weight) and quality (e.g., going from hot to cold) as well (Davies, 2016, p. 37), so we may do well to refer to this Way as the argument from change rather than motion.
I understand the argument to reason as follows:
If something is moved, it must either be moved through itself or by something else.
Something cannot be moved through itself.
Therefore, if something is moved, it is moved by something else (from 1 and 2).
If something is moved by something else, the chain of ‘causes of motion’ either regresses to infinity, or reaches a terminus in a first, unmoved mover.
The chain of causes of motion cannot regress to infinity.
Therefore, if something is moved by something else, the chain of explana- tory movers reaches a terminus in a first unmoved mover (from 4 and 5).
Therefore, if something moves, the chain of explanatory movers terminates with a first unmoved mover (from 3 and 6 by hypothetical deduction).
Something is moved.
Therefore, the chain of explanatory movers must end with a first unmoved mover (from 7 and 8).
The empirical premise here is (8): some things do in fact change in location, quantity, or quality. Aquinas argues for (2) by noting that anything that has the potential to change cannot at the same time already actually be in that changed state. But things can only go from potentiality to actuality by the action of something that is already actual (here he gives the example of fire, which is actually hot, causing other things that are potentially hot to become hot). Thus, anything that changes must do so because something external to it brings about that change. The argument for (4) requires us to distinguish per se and per accidens causal series: a series of causes that obtain simultaneously (e.g., my moving a rock by moving a stick by moving my arm), and a series of causes that exists sequentially in time (e.g., an explosion causing a landslide, forcing a town to evacuate, creating a traffic jam, etc.), respectively. Aquinas argues that an infinite per se series (also called an ‘ordered’ series) of movers is impossible because if every item in the series is merely transmitting, as it were, the causal movement from the mover above it in the chain, then if there is no first mover to serve as the ultimate source of that transmitted motion, nothing at all could ever move. For example, imagine pushing a rock with a stick. It is because I stand as the origination of the series of changes that my arm can move the stick that moves the stone – without me to act as the fundamental source of the relevant motion, nothing would move, neither the stick nor the stone. An infinite per se series of movers would be like a chandelier hanging above our heads that extends infinitely upwards with no first chain anchored to a ceiling (CITATION). It would be impossible for such a chandelier to stay in place above us. From these premises, it follows that an unmoved mover exists.
The most vulnerable premises here will be (1) and (2). As stated, premise (1) fails to list all the logically possible alternatives: there may be no cause for the motion whatsoever. In other words, Aquinas seems to take what later philosophers label a principle of sufficient reason for granted, that for any con- tingent fact x, there is an explanation or sufficient reason why x is true. But this assumption on Aquinas’ is arguably excusable. In any investigation into any aspect of the world, we generally take for granted that there is an explanation for the source of our quandary. Even if it might be complex and beyond our current ability to grasp, we still think that there is an explanation. Surely the burden of proof is on the one who wants to make some phenomenon an exception to the principle of sufficient reason, even if there are other known exceptions to it. Even if quantum phenomena, for instance, have no explanation, that fact would not entitle us to dismiss the necessity of an explanation in contexts entirely unrelated to quantum mechanics. All things being equal, we ought to assume there is an explanation, a principle that is certainly compatible with relinquishing the quest for explanation when all things are not equal (e.g., quantum mechanics).
Premise (2) faces a pretty clear counterexample: the human will. When I will to lift my hand, what causes the change in my will? Many critics of Aquinas would claim that my will moves itself and therefore premise (2) must be false. It seems to me, however, that one option in contemporary libertarian accounts of free will is that I am the direct cause of my willing to do certain things. But this reply may just bring up a new issue: if I can act as an unmoved mover in causal chains, then the conclusion of the First Way may not have much theological significance if it is compatible with there being a human unmoved mover rather than a divine one.
#2 Efficient Causation
The Second Way retraces much of the same ground as the First, this time zeroing in on efficient causation or the act of bringing something else into being (as when Michelangelo brings The David into being). Suppressing implicit premises for the sake of space, I formulate the basic logic of the Fourth Way with Pawl (2012) as follows:
There is an ordered series of efficient causes.
It is impossible for something to efficiently cause itself.
It is not possible for an ordered series of efficient causes to regress infinitely.
Therefore, there is first efficient cause.
Premise (10) expresses the empirical insight: there are efficient causes in the world that depend upon other causes in an ordered series (‘ordered’ in the per se sense defined above). It would be hard to deny this point. The argument for (11) is, in Pawl’s (2012) words, that for something to bring itself into being, it must exist logically or explanatorily prior to its own existence, a feat which is impossible. The justification for (12) is the same as that given in the First Way: without a first member to originate the series of efficient causes, nothing in the conduit of causes could ever bring anything about. From these premises, it follows that there is a first efficient cause that originates the series of efficient causes we observe, and this cause is itself uncaused.
Premise (10) presents interpretive difficulties because it is not entirely clear what examples of ‘ordered efficient causes’ Aquinas had in mind. He cannot have meant the series of, for example, my ancestors that terminates with me – for that is a per accidens causal series, not a per se one. But then, it would seem that the only clear instance of a per se series of efficient causes would be that of motion or change, a point that Pawl (2012, p. 120) makes well. If that is the case, then the Second Way collapses back into the First. Now, Davies (2010) argues that in “contrast to the First Way, the Second Way proceeds by reference to observable causality taken itself as an effect to be explained by reference to a further cause, and not from a non-causal effect to a cause” (p. 39), but it is difficult to see how this is a distinction with a difference rather than a mere verbal difference. If so, then in effect there is really only one Way cast in two different manners of speaking about the same underlying logic. One might try to absolve this charge of redundancy by casting the First Way as an argument focusing on a thing’s acquisition or loss of properties, and the Second Way as focusing on a thing’s existence (Speaks, 2006). ‘Coming into existence,’ after all, is not properly an instance of something changing or transitioning from non-existence to existence, but rather the absolute beginning of a thing. While correct, this response misses the point: unless Aquinas has a per accidens causal series in mind, where things come into and go out of being over time, then the only kind of per se causal series that anyone can think of concerns changes in quality or quantity - i.e., motion or change. There just do not seem to be any instances of the kind of causal series Aquinas needs to make the Second Way distinct from the First. Now, one might reply that surely there are per se series of causes that explain why something exists rather than changes: I exist now because my body remains intact, and my body remains intact because Earth’s gravity holds it together and physical forces keep my molecules and cells tied together, and I rely upon space itself for my continued existence. Something, ultimately, must terminate this series of explanations. But then, this chain of reasoning looks no different from the Third Way. Either way, the Second Way is redundant.
One response to this issue, for those willing to move past Aquinas’ broader metaphysical views, is to argue that certain per accidens series of efficient causes cannot extend infinitely into the past but must terminate in an uncaused, first cause, as Medieval Islamic philosophers argued and as contemporary work on the kalam cosmological argument carries forward.
#3 Contingency
In the Third Way, Aquinas builds from the insight that some things exist which could possibly fail to exist. Such things are contingent. Following Pawl (2012), we can summarize this argument to God as follows (though the ‘two stage’ idea is mine):
Stage One
If some contingent things exist, then either everything is contingent, or at least one thing is not contingent (i.e., necessary).
Some contingent things exist.
Everything cannot be contingent.
Therefore, there is at least one necessary thing.
Stage Two
If there is a necessary being, then either there is a self-existent necessary being that grounds the necessity of all other necessary beings, or the chain of necessary beings recedes infinitely.
The series of necessary beings cannot recede infinitely.
Therefore, if there is a necessary being, there is a necessary being that grounds the necessity of all other necessary beings (from 17 and 18).
Therefore, there is a necessary being that grounds the necessity of all other necessary beings (from 16 and 20).
Turn first to stage one. (14) is pretty solid, so moving on to (15), Aquinas points out that we observe things coming into and passing out of being. But for any object that comes into or goes out of being, its non-existence is obviously possible (at one time, after all, it did not exist, so that state of non-existence must surely be possible). Aquinas argues for (16) through a fairly complex chain of reasoning. For any contingent object, there must be a time when that thing does not exist. Therefore, if everything is contingent, there must be a time at which everything does not exist, or, more simply, a time when nothing exists. But if literally nothing exists at a certain point in time, it would be impossible for anything to ever come into being at a later time. After all, things only come into being if something that already exists brings them into being. But if nothing exists, then nothing exists to bring anything into being! Clearly, however, things exist now. Therefore, it is false that everything is contingent; or, equivalently, there is at least one thing that is not contingent, that is necessary. Stage two then proceeds to argue that not only must there be a necessary being, but there must also be at least one being that does not derive its necessity from anything else, a being that exists a se. Premise (18) expresses the fact that necessary beings can either be dependent or independent – that is, they can either derive their necessity from an explanatorily prior necessary being, or they can be self-existent or necessary by their own nature. But the chain of dependent necessary beings is either infinite or finite. If it is finite, then the first member of the chain of necessary dependent beings will be independent or self-existent, and it will be the ultimate explanatory ground for the necessity of all other members in the chain. As such, there is either an infinite chain of necessary, dependent beings, or at least one self-existent necessary being. The argument for (19) is the same one against infinite regression in the first two ways. Therefore, there is at least one self-existent necessary being that is the source of necessity in all other necessary beings. The Third Way leads, then, not merely to a necessary being, but to a self-existent necessary being as well.
There are two problems with this argument. First, Aquinas merely assumes, rather than argues for, the claim that if nothing exists at a time t, then nothing will exist at any time t + n. But if there can be completely uncaused beginnings, then something at t + 1, for instance, might come into being from nothing at all. At the very least, something more needs to be said about why we should think that everything that begins to exist requires a cause. Now, Aquinas could with plausibility reply in the same way as I replied to those who object to the principle of sufficient reason as it comes to fruition in the First Way: the burden of proof is on the objector to give reasons for thinking that the whole collective of contingent reality is an exception to the causal principle.
Second, as many commentators have since pointed out, Aquinas seems guilty of an illicit modal operator shift when he reasons from the claim that if there must be a time t at which any particular contingent x does not exist, then there must be a time t when no contingent things exist. We may grant that
For any particular candy bar in the world, I can eat it at a time t.
But it does not follow from (1) that therefore,
I can eat all candy bars at some time t.
I obviously cannot consume all candy bars at once; I would surely perish before completing the task! Similarly, it may well be the case that,
For any contingent thing, there is a time t at which that thing does not exist.
But it does not follow, therefore, that,
There is a time t where no contingent things exist.
Leftow (2010) bucks the conventional wisdom and defends Aquinas from the charge of quantifier shift fallacy, claiming that we can make Aquinas’ reasoning plausible if we construe the argument as something like: if every individual thing does not exist at some time t, then it is possible for there to be a time t at which nothing exists. Given an infinite past, there has been sufficient time for this outcome to materialize.
This argument is not good. First, if for any particular contingent thing, there is a time at which it does not exist, and contingent things cannot come into being without a cause, then it is metaphysically impossible for there to be a time at which all things fail to exist, and so we seem to have all the justification we need for rejecting Leftow’s modified inference to the possibility that there could be a time at which nothing existed. Second, if there are an infinite number of possible outcomes, then even infinite time would not guarantee that every possibility exists at some point. If I flip a 6-sided die an infinite number of times, then every side will show up. If I flip an infinite-sided die an infinite number of times, will every side show up? There is no way to ground an inference here. So, Leftow has a high burden of proof to make his Aquinas-revision plausible, and he has not met it.
#4 Gradation
The fourth way stems from the empirical fact that things we observe admit of gradations in their truth, goodness, and nobility. Some things are better than other things with respect to those three metrics. The argument proceeds as follows:
There are things that are more or less noble, true, and good than other things.
If things are more or less x, then there must be at least one thing that is maximally x.
Therefore, there is at least one thing that is maximally noble, true, and good (from 20 and 21).
Anything that is maximally noble, true, and good is also maximally real.
Anything that is maximally x is the cause of all else that is x.
Therefore, there is something that is maximally real (from 22 and 23).
Therefore, there is a maximally real thing that is the cause of all other real things (from 23 and 24).
As Pawl (2012, p. 123) points out, many of these premises will strike us as strange and we will accordingly be quite surprised to learn that Aquinas takes many of them as given and obvious. But we can sketch possible lines of defense for them. Focusing just on goodness, premise (20) is prima facie true – some things are better than other things. Premise (21) tells us that if we grant that point, then there must be some maximally good entity that is the standard by which these other things measure up. For if there were no such maximal standard, then it is not clear what it would even mean to say that something is more good or less good in any objective sense. The argument here is broadly Platonic. Premise (23) falls out from Aquinas’ broader metaphysic of degrees of being and the intrinsic connection between truth, goodness, and being. The basic idea seems to be that if we understand ‘truth’ to refer, not to correspondence with reality, but as an intrinsic property that objects have which is, roughly, their intelligibility or “that in virtue of which it can be grasped by intellect” (Wippel, 2000, p. 471), then something that is maximally intelligible will be maximally real. Premise (24) is not obvious, but Aquinas gives the example of fire being maximally hot and the cause of heat in all other things. In general, for any genus, the maximum exemplar in that genus is the cause of the essential features of that genus in all other members of the genus.
Given its reliance on a metaphysic that most philosophers today reject, it would serve the natural theologian to recast the Fourth Way so that it does not require commitment to an Aristotelian view of truth, being, and goodness. It seems to me that we can see the glimmerings of a modern moral argument for God in the Fourth Way. Some actions (giving to the poor) are morally better than other actions (lighting a house on fire). This difference in value, moreover, is objective. But there can only be objective differences in moral value if there is an objective moral standard against which these comparative judgements can be made (things are better the closer they approximate this standard, and worse the less they do so). One could then argue that God serves as the most plausible candidate for an objective moral standard. Of course, this reasoning leaves much of the Fourth Way behind, but it does capture and re-purpose its central observational insight: some things are better than other things.
One objection is that there is no need for an objective standard independent of the individual good things because things can be better or worse relative to one another. John can be morally superior to Mark, and both of them worse than Jane, even if there is no standard independently of these three persons by which we might judge them. To illustrate: things can be objectively bigger or smaller than other things even if there is no objective ‘big’ or ‘small.’ But that illustration demonstrates where the objection goes wrong: there are no objectively big or small objects; things are only relatively small or big. By contrast, some actions are, not merely morally better than other actions, but morally good, full stop and in an absolute sense. Removing the objective standard of goodness prevents us from making these absolute claims about goodness.
#5 Directedness in Nature
In the final Way, Aquinas draws our attention to the fact that non-intelligent objects and processes in nature tend towards their best end. Though Aquinas himself provides few examples of what he has in mind, we might point to: plants naturally grow towards sunlight, cats instinctually draw their kittens in to receive milk, bees instinctually collect nectar and produce honey for the hive, and so on. In these ways, natural processes evidence a certain directedness towards good ends that remains inexplicable unless an intelligence has acted to direct these processes towards their proper end. Pawl (2012) outlines the explicit structure of the argument as follows:
There are some natural processes that lack cognition but act for ends.
Anything that lacks cognition cannot act for an end unless an intelligent, conscious agent directs it to do so.
Therefore, there is an intelligent, conscious agent that directs some natural processes.
We have seen observations that support (28). The idea behind (29) is that nothing can tend towards its naturally best end by chance alone. Aquinas gives the famous example of the archer shooting an arrow which strikes the bull’s eye. The arrow itself has no ability to direct its actions; rather, the archer is the conscious agent that has directed the arrow to its best end (i.e., hitting the bull’s eye). From these two premises, it follows that there is at least one intelligent, conscious agent that directs certain processes in the natural world.
The most obvious objection to (29) is that advances in biology and evolutionary theory show how ‘directedness’ can arise from an impersonal, natural process. I take the implication of evolutionary theory to be that hidden in Aquinas’ reasoning is a false dilemma: either the directedness of natural objects is due to (i) random chance or (ii) intelligent oversight. But this dilemma leaves out the possibility of (iii) some level of physical necessity operating on random conditions (i.e., the theory of evolution by natural selection). If that possibility is a legitimate one, then we cannot conclude from observed coincidental alignments in nature – between behavior on the one hand and the attainment of a thing’s natural end on the other – as evidence for God. Now, Pawl (2012, p. 124) points to processes outside the scope of Darwinian evolution that still demonstrate means to ends behavior, like an electron naturally attracting positively charged particles. But the problem is that in cases like these, it is difficult to prove that the specified result (electrons attracting protons) is the best end of the object in question (the electron). Electrons, I should think, are not much affected whether they attract protons or not! More importantly, we have physical, scientific explanations for the attraction between electron and proton that seem perfectly sufficient and do not require recourse to an organizing intelligence.
I see two ‘take aways’ from Aquinas’ natural theological project. First, even if Aquinas’ case fails, natural theology is still a worthwhile pursuit. Considering the obstacles facing Aquinas, some philosophers may conclude that arguments for God are always and everywhere outmoded (that word reminds me of Edna Mode). But that conclusion is too hasty. I think many theistic arguments are worth considering if we put in the work to refurbish them in light of contemporary philosophy. Just as poor science in times gone by does not undermine contemporary science, neither does the failure of past arguments for God show that present attempts at natural theology are misguided or doomed to failure. Indeed, one conclusion from the Five Ways is that philosophy, like science, makes progress. Now, whether there are successful arguments for God is a different question, but my point is merely that it is a question that is still worth debating.
Second, Aquinas strikes the right balance between empirical observation and a priori reasoning. Philosophy - I think, anyway - ought to engage in both, and contemporary debate in the philosophy of religion has taken that insight to heart. Some of the most-discussed arguments for God stem from contemporary scientific investigation, the origin of the universe and cosmic fine-tuning being the most obvious examples. Still, philosophers are not scientists, and they can avail themselves of a broader philosophical explanatory toolkit. Like Aquinas, philosophers today ought to affirm the autonomy of philosophical investigation, all the while stay abreast of contemporary scientific evidence.
Davies, B. (2014). Thomas Aquinas’s Summa Theologiae: A Guide and Commentary. Oxford University Press.
Kenny, A. (1969). The Five Ways. London: Routledge & Kegan.
Leftow, B. (2010). Arguments for God’s existence. In The Cambridge History of Medieval Philosophy. Vol. 2, ed. by Robert Pasnau with Christina Van Dyke, Cambridge: Cambridge University Press. 735-48.
Pawl, T. (2012). The five Ways. In The Oxford Handbook of Aquinas, ed. by Brian Davies and Eleonore Stump, Oxford: Oxford University Press. 115-31.
Speaks, J. (2006). Aquinas’s first and second ways. Lecture notes. https://www.3.nd.edu/jspeaks/courses/mcgill/201/aquinas-cosmological-argument.html.
Wippel, J. (2000). The Metaphysical Thought of Thomas Aquinas, Washington D.C.: The Catholic University of America Press.