Saturday, 20 January 2018

Five Proofs Critique: The Aristotelian Proof

This next entry discusses Feser’s Aristotelian proof, the argument with which he opens the book, as well as a wider problem with Aristotelian accounts of agency and explanation. Central to this discussion is a certain account of free will mentioned in the last installment.

Before discussing the argument itself it should be remarked that Feser’s presentation of the powers theory of causation as a way to answer Parmenides is ambiguous in that it slips from making a common-sense case for a powers theory to endorsing a specifically Aristotelian or Thomist account of powers and ontological composition. Many powers theorists prefer a pan-dispostionalist account on which there are no actual or ‘categorical’ properties. Even from the claim that substances have actual properties/categorical properties, active potencies (powers or capacities) and passive potencies (dispositions) it does not follow immediately that said substances are a composite of prime matter and actuality. Let us arrange these in a tier system:

1a. Objects possess powers and dispositional properties.
1b. Objects possess powers, dispositional properties and categorical properties.
1c. Objects are composites of Act and Potency.

The last position will require a far greater amount of technical argument to support than the former. But let us ignore these concerns for the present and appraise the argument on its own merits.

The Aristotelian argument is near identical to Thomas Aquinas’ First Way.  One argues from the existence of change, the assay of change as the actualisation of a potency, and the fact that change requires a changer to the existence of an unchanged changer, a being which is pure actuality. Additional Aristotelian metaphysical premises are then appealed to in order to prove said being has all the necessary divine attributes.

My commentary will focus on the distinctive premise of the argument, the claim that everything that is changed is changed by another. This claim is sometimes referred as the principle of motion in line with the older tendency to translate Aristotle’s Greek term for change as ‘motion’ (hence the famous epithet ‘unmoved mover’). Needless to say Feser accepts this principle and claims it is synonymous with the principle of causality. The objection or at least concern I will raise dates back at least to that other ‘greatest medieval philosopher’, Duns Scotus, and is suggested by Aristotle’s own discussions of the sense in which animals can be said to be self-movers. Put simply there is a prima facie tension between the claim everything that is changed is changed by another and the apparent fact that agents possess free will in the true libertarian sense1.

Throughout the course of the essay it will be assumed that agents do indeed possess free will. Aside from it being a pre-philosophical assumption we operate on every day of our lives there are two further reasons for this. The first is that many philosophers are impressed by the Consequence Argument to the effect that determinism is incompatible with agents being responsible for their actions. The second is that, as was seen in the previous instalment on the Leibnizian cosmological argument, a certain account of free will is the most plausible candidate for resolving pressing concerns with the Principle of Sufficient Reason. Proponents of the Aristotelian argument are free to bite the bullet and accept determinism straight off but as free will is such a fundamental belief most philosophers would rather discard the principle of change than relinquish it.

There are a number of different ways of defining libertarian free will but they normally involve the following premises. They are here given in universalised forms, though one will remain open to the prospect of their applying differently to different species of acts (plausibly different acts have different conditions under which they count as free):

1. An act at time t1 is free if and only if the state of the universe (covering both physical and psychological states) at all times before time t1 are insufficient to determine that the agent would perform said act.
2. An act is free if and only if it is possible for the agent to do otherwise.

The former amounts to a denial of causal determinism (one can modify it slightly if one wishes to cover the freedom of the angels and other timeless free agents) whilst the second is the famous Principle of Alternative Possibilities, a criterion the Subtle Doctor himself proposed2.

Given that Aristotelianism is a faculty psychology and places a strong emphasis on the role of the intellect in action the term ‘act’ used in the above criterion might be too vague. To make discussion easier I shall confine myself to one specific kind of purported free act that is acts of choice. The concern is that the claim integral to the Aristotelian argument, that everything that is changed is changed by another, is incompatible with the second if not both of these principles. More so if free will is impossible then either determinism is true or there exists genuinely random uncaused events, something which is ruled out by the strong PSR.

The critic might start by observing that no being can change to perform an action (for example the raising of a hand as an instance of localised change) unless it was part of an ongoing chain of changers that only finds its termination in God. This implies however that the action change was brought about by another changer acting on the agent. If so then the agent was changed to perform the action by something beyond itself.

‘So what?’ The Aristotelian might respond. Of course certain processes of change, both internal and external, are required for the agent to perform said act, indeed it would be nonsensical to assume otherwise: without the on-going processes of respiration, metabolism and neural firings as well as all the relevant beliefs and perceptual functions the agent would not even remain in existence let alone the rationally aware state required for a free act. A purely immaterial being like an angel might require no other causal factors but as partially embodied beings we do. Furthermore there is nothing in the least bit troubling for the libertarian in this. The first criterion only requires that all the prior causal factors do not constitute a sufficient condition for the agent’s acting freely, whilst allowing that it might (and should) still constitute a necessary one.

Unfortunately there is a deeper problem, one which pertains to the second criterion and is of particular concern to the Thomist given their intellectualist psychology.  Let us say an agent is faced with a choice between two options, A and B. According to the principle of change whichever choice the agent makes is the result of being changed by an external changer.

‘You’re making the same mistake!’ The Aristotelian protests, ‘Yes, the principle of change entails that the agent choose either A or B as the result of an external changer, but it does not determine which of the two they will choose. To put it another way the agent is changed or caused to [choose A or B] not changed to choose [A] or changed to choose [B]. You are probably thinking of the question ‘Why did the agent choose A instead of B’. This is not a question about causation or change though but about contrastive explanation.  Here you already have a satisfactory answer: the agent’s choosing A instead of B is explained by their free choice just as their choosing B over A would have been’.

This pre-emptive response is coherent but has the somewhat odd consequence of committing one to disjunctive causation (specifically disjunctive effects). In all fairness Aristotelianism, with its commitment to substance as opposed to event causation, is singularly well-placed to meet the biggest objection to disjunctive causation, which is that there are no disjunctive events in nature3. Feser himself is at least open to probabilistic causation which is close in nature to disjunctive causation.  One might wonder whether this really captures the nature of action though—intuitively one’s action is not to choose one or the other but to choose a specific option; it is this specific act, not some higher genus of act, which does or does not require an external changer. Nevertheless if change or causation does not require a contrastive element and thus the relation at work in the above is instead explanation then the Aristotelian can meet the challenge, at least with regard to external factors.

The greater challenge lies in reconciling Thomistic psychology with free will. It is frequently alleged that the ‘simplistic’ intellectualist character of Thomas’ eudaimonistic theory of value leads to intellectual determinism i.e. the will is determined by the intellect the contents of which are determined by the external environment. If the intellect’s judgement always inclines one to the greater good then one no more chose the actions one will undertake than one ‘chose’ the results of a mathematical proof.

The Thomist can respond with some justification that even if the two common libertarian criteria given above are necessary for a workable theory of free will they are in themselves insufficient for a theory of human action. As well as giving an account of the will, free or otherwise, such a theory should cover motivations and value; without these the will appears aimless. The point of life is to seek the good, for the good is of its nature what ought to be sought; agents might be free to will all sorts of proximate goods, perceived or otherwise, but reason is of its essence orientated towards goodness. This follows from the nature of rationality, which is to seek truth, a transcendental property which is ultimately convertible with goodness and being. More so: the ultimate good, the only good that is truly good in itself, is union with God, something a rational being cannot help but will for it is the telos of existence itself.  

Theological concerns should also make us think twice about accepting those two points as the ultimate criterion of free will. Since Anselm it has been a dictum of classical natural theology that God is both more powerful and more free in virtue of not being able to do certain things i.e. will evil. The inability to do evil or fall prey to temptation is not a lack of freedom but an abundance of it. Quentin Smith puts the issue like this: Freedom is considered a good thing, and so God, the perfect being, must possess freedom to the maximum degree, yet God is necessarily good and thus will not make bad choices. Therefore perfect freedom must be compatible with not being able to make bad choices4.

There is much that is important in the Thomists contention. Contemporary philosophers, both libertarian and compatibilist, have criticised simplistic libertarian accounts consisting solely of indeterminism coupled with the principle of alternative possibilities for failing to capture the goal-directed character of the will. To be free means not only to be able to do otherwise but to do so in accord with the good and the true. Likewise any broadly religious theory of human nature must give some account of the role of the ultimate in human motivation. 

Let us review the account of free will appealed to in the course of the PSR Cosmological Argument. Defeating the modal collapse and no best possible worlds problems requires that, at the bare minimum, God has the capacity to choose otherwise than one actually does when faced with two or more comeasurable goods (call such scenarios Buridan cases). For human actions to be free (and thus self-explanatory) it requires that they not be determined by prior causal history and that in at least some cases it is open to the agent to act otherwise. The Thomist will protest this latter, claiming that there will be a great many cases where one could in principle deduce what choices an ideally rational agent would make based on their psychological states and thus prior causal history. This does not mean that some free choices are not undetermined and do not involve alternate possibilities only that the remit for what qualifies as a free action must be wider than those two. It is however the possibility of such cases which is of interest to us qua natural theologians; if Thomism can furnish such, it can avoid concerns with the PSR and reasons to doubt the principle of change.

It is in the very least doubtful whether Thomism can provide such an account. The problem is not that Thomism furnishes too weak a libertarian account of free will but that historically Thomists have not given a libertarian account at all, opting for one that sounds more like compatibilism5. This I suspect is the real reason why Feser does not appeal to libertarian free choices as self-explanatory in order to block the BCCF objection to the strong PSR. As far as I know he has never commented extensively on the issue of free will but on the occasions where it has come up he has defined it negatively as absence of external constraints within the created order.6 Such language is in line with the traditional Thomist tendency to define freedom as ‘the capacity to act in accord with one’s nature’ or ‘to pursue the good unimpeded’, something which no doubt stems from Thomas himself referring to the voluntary as ‘that which knowingly moved by an intrinsic principle to an end’. 

Just because an account defines one of its terms in a different way from the normal usage it does not mean that account necessarily lacks what that term normally refers to. For instance Thomists have an annoying habit of using the term ‘necessary being’ to refer to a being that exists omnitemporally though their ontology also includes beings which are necessary in the normal sense i.e. the non-existence of which is impossible. It may be that Thomism is sufficiently libertarian for our needs, only our criterion of libertarianism simply was not so relevant to the free will debate of his day.   

An effective way to test this is to look at how Thomists handle what have earlier been referred to as Buridan cases, cases where the agent chooses between two co-measurable goods. On such an occasion there is no external value-based reason for the intellect to move the will towards one or the other alternative. Thomas himself references just such a case where he emphatically denies that the agent wills out of necessity (Summa Theologiae, Prima Secundae Partis Q.13, Art6). He claims that in cases where the agent is presented with ‘two things proposed as good under one aspect nothing hinders us from considering in one of them some particular point of superiority, so that the will has a bent towards that one rather than towards the other’. Unfortunately this response does nothing to clarify the problem, indeed the qualification that the two things are ‘equal under one aspect’ seems like an attempt at evasion, since it is precisely the two options being perfectly equal in all respects, which gives rise to the paralysis of the will in the first place.   

Furthermore as the Scotist scholar Thomas Williams points out Thomists are faced with a higher order variant of the same problem—how does one decide which alternative ‘particular aspect’, which criterion to view the alternatives? If the ground for the alternative criterion lies in some difference in value between the objects (say for instance both objects of choice are equally good for health but one more aesthetically pleasing) then Thomas has just ducked the question; if, on the other hand, it lies in something unrelated to the objects and intrinsic to the agent’s psychology (say one of the objects has positive Proustian associations) then one is back at psychological determinism. If ultimately there is no further reason why the agent opts for one criterion over another then one does have fully-fledged freedom but why not appeal to this primitive freedom in the context of the original choice? 

Thomists have interpreted the above passages in a number of ways.  Some e.g. have interpreted the above in the full-blooded libertarian sense to mean that an agent has alternate possibilities open to them when faced with any choice between finite goods no matter how disproportionate7. Others, for instance Eleonore Stump, are keen to interpret Thomism as rejecting the principle of alternative possibilities in favour of an account which grounds freedom in the ability to follow one’s rational nature free from external constraint. When appraising Aquinas' system as a whole one gets the impression that it slides towards compatibilism no matter how the saint himself would have wished to avoid it (for what it is worth a similar controversy rages over the status of Aristotle’s view on free will). At best the account might leave room to be fleshed out in a fully libertarian manner.

From herein I shall return to discussion of the general Aristotelian argument. My goal in discussing the problem of free will on the Thomist account is to stress that if Thomism in particular, or the ontology behind the Aristotelian argument in general, commits one to a conclusion that is incompatible with strong PSR then that is reason for one to reject it. As philosophers our commitment to ultimate rationality goes deeper than our commitment to a specific account of causation.

But maybe the fate of the Aristotelian argument is not tied to that of the principle of change. If reconciliatory overtures to libertarianism fail and free agents are self-changers there may still be a way to salvage the argument by appealing to weaker premises. What the objection establishes is that only a free agent can be a self-changer. Surely it is possible to have change without contingent agents though. If this is so then one can make the case that there are possible worlds where the Aristotelian argument does go through. Consider:

1. Change is instigated by an agent.
2. Change is possible if contingent agents did not exist i.e. a world without contingent agents is possible.

Which leads to:

3. Therefore there possibly exists an agent which is not contingent i.e. there is a possible world containing a necessary agent.
4. A necessary being is one which exists in all possible worlds (possible world semantic definition of necessary being).
5. If a necessary being exists in all possible worlds then it exists in the actual world.

If a world without contingent agents and with change (given that change includes substantial change i.e. substances coming into existence, and mere continuance in existence any world with non-agential contingent beings satisfies this description), then the being reached by the conclusion of the Aristotelian argument is possible. This being is a necessary being, and, given the S5 axiom, a necessary being is one which must exist in all possible worlds including our own.

Of course this is a parallel to the modal ontological argument. The modalised Aristotelian argument does have an advantage over the modal ontological argument in that its possibility premise is plausibly supported by subtraction (a non-agent world could be just like this world minus some of its contents—most people would admit that it is possible for all contingent agents to go out of existence). 

This amendment is unlikely to appeal to the followers of Thomas Aquinas though, as most Thomists are apt to suffer from their own variation on what Richard Gale playfully called ‘Extreme Modal Intuition Deficiency Syndrome’8. In their case it amounts to a refusal or at least an extreme reluctance to admit something is possible until it has been proven to be actual9. Feser comes clean about this problem during his discussion of David Schrader's anti-theistic argument in Chapter 6. Since there does not appear to be a reason for it beyond possible Imagist elements in Thomist epistemology there is not much more I can say on the matter.

Finally a comment on the scope of the Aristotelian Argument. Feser and other Thomists define change as the actualisation of a potency. In what manner does change differ from (efficient) causation? True, when we talk of causation we are apt to think of entities being caused in the sense of being caused to exist, but that is only a particular type of causation10. All four of the Aristotelian forms of change are instance of efficient causation: the fire causes the pot to grow hot, the bird causes the apple to fall from the tree to the ground, the animal’s food items cause it to gain body mass, and the neutron’s colliding with the atom causes the latter to break into subatomic particles. One can see this more clearly if, instead of talking about change of a substance, one talks of the coming to be and passing away of that substance’s accident tropes. The tendency to think of causation as what Aristotelians call substantial change i.e. the coming to be or passing away of a substance, might help explain our qualms about supposed instances of self-instigating causation: although it is nonsense to talk of a substance’s causing its own existence it is not so evidently contradictory to speak of something’s being the cause of its own change in other respects i.e. its acquisition of new accidental properties11

If it is the case that change and efficient causation are identical then one might conclude that the Aristotelian argument and the First Way are unnecessary, for one can reach the same conclusions by appealing to a much weaker formulation of the same basic argument. The claim that all instances of change are instances of efficient causation contains the more modest premise on which Thomas' Second Way is based, that is, ‘every substantial change requires a changer’ or in more accessible language ‘every being that comes to exist has a cause’. This claim, with its denial of substances just happening to be there or popping into existence for no reason, has as strong a pre-philosophical justification as one could wish for. Parmenides’ famous maxim nihil ex nihilo is a safer bet for a theistic argument than Aristotle’s ‘everything that is changing is changed by something’.

By now the reader might have detected a haunting similarity between the two arguments under discussion i.e. the Aristotelian argument, which covers all property instantiation including both accidental and relational, and its ‘weaker’ derivate, the Second Way, which only covers the instantiation of substance properties, and a certain other pair of ‘strong’ and ‘weak’ arguments discussed previously. The Aristotelian argument and the Second Way of course parallel the strong and weak formulations of the PSR cosmological argument. This is to be expected—after all causation is often thought of as a kind of physical parallel to explanation, the main difference being explanation can account for the status of necessary beings, including casually impotent platonic abstracta—but the fact that the strong PSR can account for Libertarian free will and the strong causal principle cannot suggests that the Aristotelian-Thomistic account of causation is lacking something.

The writer would like to thank Cyrus Pacquin and the members of the Thomist Discussion Group for their helpful input and suggestions.


1. For simplicity's sake I will not touch on questions concerning the relationship between free will and divine providence. I am assuming all parties will accept that the two are compatible even if they disagree exactly how this is so.

2. For an excellent overview of Scotus’ account of free will with an emphasis on the principle of alternate possibilities see Thomas Williams’ ‘The Libertarian Foundations of Scotus's Moral Philosophy’. I have drawn extensively from this source in my discussion of Thomas’ alleged Compatiblism.

3. Discussions of disjunctivism in causation alternate between disjunctions as causes and disjunctions as effects. For accounts of both albeit in the language of event causation see Carolina Sartorio’s ‘Disjunctive Causes’ and Roberta Ballarin’s ‘Disjunctive Effects and the Logic of Causation’.

4. For this argument please see pp 149-157 of Quentin Smith’s Ethical and Religious Thought in Analytical Philosophy of Language.

5. For a discussion of Thomas' account of free will versus modern Libertarian accounts see pages 277 to 306 of Eleonore Stump’s Aquinas. She gives a compatibilist tinged reading complete with appeal to Frankfurt counterexamples to the principle of alternate possibilities.

6. See his brief discussion of free agents’ behaviour in the video Faith and Reason 2-16-11. They occur at around the 130 minute mark.

7. For these see Kevin M. Staley’s ‘Aquinas: Compatibilist or Libertarian?’ and Kelly Gallagher’s Can Libertarianism or Compatibilism Capture Aquinas' View on the Will? Gallagher cites Garrigou Lagrange as endorsing her interpretation. Staley admits that the text itself is unclear.

8. This claim was in response to Peter van Inwagen’s modal scepticism.  Sufferers have a tendency to disregard modal claims even if they are constructed out of modal intuitions we otherwise rely on. At the other extreme one has those who proudly declare without further elaboration that conceivability entails indefeasible possibility and that they can ‘conceive’ all sorts of things just because these ‘conceptions’ do not violate narrow logical necessity. Ironically given their proclaimed anti-metaphysical stance Humeans and empiricists are most frequent victims of this complaint.

9. It should be noted that Feser does also give an explanation—Brian Leftow’s explanation it has to be said—as to why the conceivable and thus prima facie possible scenario Schrader appeals to does not represent the possibility he claims. Further instances of this EMIDS-induced tension can be found in his account of the conceivability argument for dualism in Philosophy of Mind and in various blog entries discussing Descartes’ ‘clear and distinct’ perception argument.

10. A reason for the confusion might be that talk of substantial change appears to imply that (as Aristotle himself thought) there is a something which is being changed from one substance to another, some substantial substratum like the dreaded prime matter. This fits badly with the specifically Abrahamic notion of creation ex nihilo, as well as the genesis of spiritual beings like the angels. Indeed talk of causation as the actualisation of a potency only makes sense if one takes the potency in question to be what Scotus called objective or logical potency, or that an essence stands in potency to an act of existence.

11. Despite the pejorative claims of many determinists e.g. Nietzsche and Galen Strawson, self-change is not a genuine case of causa sui. A substance might cause its new accident tropes to exist but this is not cyclical; indeed it would only be so if the accident tropes that had just came into existence somehow caused themselves to exist. It also does not escape one’s attention that said determinists promptly cease being realists about even much weaker realist accounts of causation when they come up in the context of cosmological arguments. 


Ballarin, Roberta (2014). ‘Disjunctive Effects and the Logic of Causation’. The British Journal for the Philosophy of Science, Volume 65, Issue 1.
Feser, Edward (2014). Scholastic Metaphysics: A Contemporary Introduction. Ignatius Press.
Feser, Edward (2017). Five Proofs of the Existence of God. Ignatius Press.
Gale, Richard (2004). God and Metaphysics. Prometheus Book.
Gallagher, Kelly (2014). Can Libertarianism or Compatibilism Capture Aquinas' View on the Will? University of Arkansas. Accessible here.
Kane, Robert (2014). A Contemporary Introduction to Free Will. Oxford University Press.
Rowe, William (1975). The Cosmological Argument. Princeton University Press.
Sartorio, Carolina (2006). ‘Disjunctive Causes’. Journal of Philosophy 103. Accessible here.
Smith, Quentin (2004). Ethical and Religious Thought in Analytical Philosophy of Language. Yale University Press.
Staley, Kevin M (2005). ‘Aquinas: Compatibilist or Libertarian?' The Saint Anselm Journal. Accessible here.
Stump, Eleonore (2003). Aquinas. Routledge.
Williams, Thomas (1998). ‘The Libertarian Foundations of Scotus's Moral Philosophy’. The Thomist. Accessible here.


  1. Quick note on a small error (feel free to not publish this comment, since it adds nothing substantive): The Ballarin, Roberta (2104) reference appears to have an inaccurate date.

  2. How do you define the will. Not *freedom* of the will, but the will itself. What is the will to you?