Wednesday, 27 February 2013

Second Post on Aquinas's First Way: the Impossibility of an Infinite Regress of Moved Movers



          
               Having argued that if anything is moved it is moved by something else, Aquinas now argues that it is not possible that this continue indefinitely. 

But this cannot go on to infinity, because then there would be no first mover, and, consequently, no other mover; seeing that subsequent movers move only inasmuch as they are put in motion by the first mover; as the staff moves only because it is put in motion by the hand. Therefore it is necessary to arrive at a first mover, put in motion by no other; and this everyone understands to be God. 

Read uncharitably, this part of Aquinas's argument might seem nothing more than question-begging. That there is no First Mover is precisely the thing his opponent would question. Nor has Aquinas, the objector may continue, provided a reason not to think that the series of movers could (possibly) extend to infinity. But there is, I think, a more plausible way in which one can read him. There is a distinction that Aquinas (and later Duns Scotus) draws between a series of efficient causes ordered per accidens and a like series ordered per se. I now try to describe that distinction in some detail.
             
            Now, the main point behind the distinction, one should note is the following:
          
(NIR): for any series of efficient causes ordered per se, an infinite regress of causes is not possible.

Though Aquinas elaborates on the distinction the point of which is to prove NIR somewhat in his Summa Contra Gentiles, a more detailed discussion of it is found in the writings of John Duns Scotus, Aquinas's near contemporary.
            In his own attempt at a proof for the existence of God, from his De Primo Principio, Scotus provides three criteria for distinguishing a series ordered per accidens versus a series ordered per se. To begin with the latter, a series ordered per se is (1) such that all the efficient causes in the series are either dependent (with respect to their causal efficacy) on the causes that precede them in the series, being first in the series in the sense of depending on no other causes in the series for their act of causing, or both are depended on and depend on other causes in the series. Second (2), in a series ordered per se, the type of causality of at least one of the efficient causes is of a higher type and rank, "inasmuch as the higher cause is more perfect." Scotus also argues that (2) is a consequence of (1) because no cause, in its exercise of its causal power, is essentially dependent on a cause of the same type as itself. The final difference (3) is that all of the efficient causes in a series ordered per se are simultaneously required for the series to have causal efficacy. In this last connection, Scotus is not explicit on whether he thinks (3) also follows in consequence of (1), but it seems he is on good grounds in claiming so. For, if the first cause in the series fails to exercise its causal efficacy precisely at the same time as the second cause is required to exercise its causal efficacy, it (the second cause) will have no causal efficacy with respect to what it requires for bringing about its effect.
            The staple medieval example of a series ordered per se is a hand's moving a staff to move a stone. In keeping with (2), the hand is a higher type of cause from the staff because the hand (or, more accurately, the mind controlling the hand) can (in some sense at least) by itself initiate movement or causal activity in the series, whereas the staff cannot. Since the staff depends on the hand precisely in the staff's act of causing, moreover, the example also satisfies (1). Finally, arriving at (3), the staff and the hand must exercise their causal activity at the same time, and therefore must both be present such that the hand can act with the staff. The relations that obtain in virtue of (1) and (2), one might note, are asymmetrical: if a causes b to act, it does not follow that b causes a to act. They are also transitive: because if a causes b to act and b causes c to act, it also follows that a causes c to act. One might object to the claim that the relations that obtain between a and b in virtue of (1) and (2) are that of asymmetry and transivity, on the basis that a is also dependent on b in a's act of causing. My reply is that this response succeeds only if one equivocates between a and b in the sense of (1) and (2), on the one hand, and, on the other hand, a and b in the distinct sense implied by the relations that obtain between a and b in virtue of (3). For the relation between a and b in sense (3) is indeed symmetrical, and in this third sense it is true both that b depends on a and that a depends on b. But this sense, I hope is clear, is perfectly distinct from that logically implied by senses (1) and (2).
            In contrast to a series of efficient causes ordered per se, on the other hand, there is also a series of efficient causes ordered per accidens. Having defined a series ordered per se in terms of (1) through (3), a simple and clear way of explaining a series ordered per accidens is as merely a series in which (1) through (3) do not obtain. The standard example here is that of Abraham's begetting Isaac and Isaac's begetting Jacob. As his existence is concerned, so the example goes, Isaac depends on Abraham's having begotten him some time in the past. As it concerns his act of begetting Jacob, however, Isaac does not, in any active sense, depend on Abraham. He can act so to beget Jacob even if Abraham is no longer alive.
            Each member that is part of a series of efficient causes ordered per accidens, therefore, is independent of any prior member so far as its causal efficacy is concerned. In consequence, the time that separates each act of causality in the series is also irrelevant insofar as each agent's causal efficacy is concerned. That is, the causes need not exist simultaneously for the series as a whole to have its causal efficacy. For this reason, then, it is possible that a series of causes ordered per accidens stretch back infinitely into the past.
           
            The point of my describing this distinction between a series ordered per se and a series ordered per accidens should now be fairly clear. For Aquinas to rule out the possibility of an infinite regress with respect to motion, he needs to argue that the relevant classification for the series of efficient causes contained in the series in question is ordered per se and not per accidens. In this regard, one reasonable starting place for arguing this conclusion, perhaps, is to see whether the example of the hand, staff, and stone, in terms of their motion, can be explained in terms of a series ordered per accidens. That this is an appropriate starting point, moreover, seems plausible in light of the fact that this very illustration matches up with the example Aquinas himself uses, as the very end of the last passage I quoted from the Summa reveals. In any case, can the example be understood in a way that permits an infinite regress to occur? According to William Rowe, it cannot.
            That is, no it cannot, Rowe claims, if one wishes to understand why the stone-moving activity is now going on.[1] One does not arrive at an explanation of the stick's present movement, in other words, if one refuses to postulate something more than intermediate causes. For, otherwise, he continues, the series will proceed to infinity and one will fail to arrive at any explanation for the motion going on with respect to the stick right at this moment.
            But is it really necessary, Rowe then asks, that one has to posit that, for anything x such that x is an effect, x is caused by something y? Rowe is pointing out, in other words, that Aquinas's argument seems, at this point at least, to depend on some version of the principle of sufficient reason (PSR). Very roughly, the version of PSR that Rowe is suggesting that Aquinas needs is something like: "for any movement or change x, there is a fact or an explanation y such that y explains x." While Rowe is indeed correct to point out that Aquinas's argument does require some version of PSR in order to be successful, this may be somewhat besides the point.
            For given the concept of change as I have argued that Aquinas understands it, change is a process or an event that simply occurs necessarily (in the sense of natural necessity) in light of the fact that there is an agent that is actually F and a suitably disposed patient that is potentially F such that if, in these circumstances, the agent can bring about F in the patient, then it will necessarily bring about F in the patient. The only circumstances in which the agent will fail to bring about F, moreover, is if the patient is improperly disposed to receiving F, e.g., the patient is either too far away from the agent or the agent's natural power to produce F or the patient's natural power to receive F is otherwise obstructed. Thus, assuming Aquinas's account of motion is coherent, for any movement or change x, there will be a natural explanation y for why x takes place. Rowe might protest this move, insisting that Aquinas needs something stronger than natural necessity, but it is not clear why this would be the case, since all that Aquinas, by Rowe's own admission, seems to need is an explanation for every change or movement that takes place; there was never any specification on Rowe's part for a full explanation or a necessary (in some stronger sense of necessity than natural) explanation. So, in other words, the version of PSR that Rowe seems to believe Aquinas needs comes "for free" along with the Aristotelian metaphysics that Aquinas accepts. The only other way, so far as I can see, for Rowe to get around this difficulty is to attack Aquinas's premise that anything that is changing is being changed. And I have already given some reasons for thinking that his attack fails on this front.
Are Infinite Regresses Really Impossible?
            But assuming that Rowe's objection is not sufficient to block the First Way, and assuming that a series ordered per se can explain certain sorts of motion, a question remains: why exactly is a series ordered per se necessary to prevent an infinite regress of motion? Take the example of the hand, the staff, and the stone. Now, it cannot be the case that the staff is a necessary condition for the stone's being moved, because, for one, one could move the stone with something besides a staff, and, more importantly, someone could move the stone without the staff at all. So, at most, Aquinas can say that the hand and the staff moving the stone are sufficient conditions for moving it and not necessary conditions. This raises an important question: does the hand's moving the stone still count as a series ordered per se? In terms of the conditions necessary for something's counting as a series ordered per se, it seems true that the hand is a higher type of cause than the stone, in the sense that the hand can initiate movement and the stone cannot. It is also true that the hand and the stone must both be present at the same time. What is not necessarily true is that the stone depends on the hand precisely in its act of causing motion, because it is not necessary that the stone produces further motion in another thing. So, apparently, the hand alone moving the stone is not necessarily part of a series ordered per se. Hence, it cannot be true that all cases of motion involve such an ordered series. But this conclusion may be too quick.
            For, here, one must recall that the illustration was merely an analogy. For, even with respect to the hand, the hand will be moved by the mind of the person moving the hand, and the mind of the person will be moved by the movement of neurons in the person's brain, etc. And all of this, it seems reasonable to think, must take place in a finite period of time, because all motion, in the sense of something moving from on state of potency to a state of actuality and vice versa, takes place in a finite period. Moreover, even if one were to argue that some of the causes in the series stretch back into an infinite period of time, it is clear that some of the causes that cause movement in someone's brain must be simultaneous with this movement, because the brain/mind's moving the hand must be simultaneous with the hand's moving the staff, and also simultaneous with the staff's moving the stone.
            But, clearly, it seems, the series of movers involving the mind, hand, etc. must come to an end, nor do any of these items seem to have the power to initiate movement, as this would imply that they move themselves from potency to actuality, and this is not possible. And this is all the argument needs to generate a potential regress stretching back until one arrives at a first mover. Moreover, even if one were to concede that some series of motion are not ordered per se, this would still not provide a satisfactory explanation for those series that are, in fact, ordered per se. And this is all Aquinas would seem to need for proving that some series must have a first mover. Thus, whether Aquinas can succeed in showing that the first mover must be God is a question I leave for another time. But it seems clear that, in arriving at this point, he can do a much better job than many commentators seem to think.

No comments:

Post a Comment