Wednesday, September 12, 2018

The First Way and the Fire Example

Michael Almeida, in his recent book, Cosmological Arguments (which you can, for a short time, read for free online), on Aquinas's argument in the First Way against self-causation:

The argument here appears to be the following. If y is actually F and potentially G, and x is the cause of y actualizing G, then x must be G. So, if y is the cause of y actualizing G, then y must already be G. Therefore, if y is the cause of y actualizing G, then y is both actually G and potentially G, and that is impossible. Therefore nothing causes itself to move or undergo change.

Aquinas’s argument against self-causation is unfortunate. It entails that anything x that causes y to be some property G must itself be G. Thus, taken at his literal word, the fire that causes the wood to be hot must itself be hot. But it is obvious that something might cause an object to become green without actually being green. Copper undergoes chemical reactions that cause the metal to change to a pale green. But the cause is not pale green. Similarly corroding iron changes from silver to orange-red. The cause of this change is not itself orange-red. Examples are easy to multiply.

Analytic philosophers have been interpreting the argument this way ever since Anthony Kenny (who should have known better) interpreted it this way, but there is good reason to regard it as a misinterpretation. First some rough working through the Latin, without worrying too much about finer points:

Nihil enim movetur, nisi secundum quod est in potentia ad illud ad quod movetur, movet autem aliquid secundum quod est actu. Movere enim nihil aliud est quam educere aliquid de potentia in actum, de potentia autem non potest aliquid reduci in actum, nisi per aliquod ens in actu, sicut calidum in actu, ut ignis, facit lignum, quod est calidum in potentia, esse actu calidum, et per hoc movet et alterat ipsum. Non autem est possibile ut idem sit simul in actu et potentia secundum idem, sed solum secundum diversa, quod enim est calidum in actu, non potest simul esse calidum in potentia, sed est simul frigidum in potentia. Impossibile est ergo quod, secundum idem et eodem modo, aliquid sit movens et motum, vel quod moveat seipsum.

A bit clunkily: "Nothing is changed except inasmuch as it is potential to that to which it is changed; but it changes something inasmuch as it is actual. For to change is nothing other than drawing forth something from potency into act, but something cannot be reduced from potency into act, except through some actual being; just as the actually hot (like fire) makes wood (which is potentially hot) to be actually hot, and thereby moves and alters it. For it is not possible for the same thing to be at once actual and potential inasmuch as it is the same, but only inasmuch as it is different, for what is actually hot cannot at the same time be potentially hot, but is at the same time potentially cold. It is therefore impossible that, in one and the same way, something could be moving and moved, or that something move itself." Note, incidentally, that one could argue that the per aliquod ens in actu could be read as already implying that we are talking about another thing, and not just the generic 'some thing'.

Aquinas explicitly says -- twice, even -- that the point is that the cause of change must be actual. What Aquinas is arguing in context is that everything moved is moved by another; what he is doing here is specifying what that means in terms of actuality and potentiality, which, of course, are more fundamental than change. It is not correctly interpreted as "If y is actually F and potentially G, and x is the cause of y actualizing G, then x must be G"; what it actually says is "If x is the cause of y, which is potentially G, actualizing G, then x must be actual."

Indeed, the interpretation that takes him as saying that x must be G is interpreting him as proposing a claim that implies the claim, "All causation is univocal." We know that Aquinas does not think that all causation is univocal; that it is not is said and assumed in a wide number of passages. It's not as if it's something he rarely talks about, and has given little thought to. Thus the interpretation in and of itself should raise warning flags.

However, Almeida considers a suggestion from Craig that is essentially the right interpretation (although Craig underplays it as a charitable interpretation, it is, as I note above, what Aquinas explicitly says) and rejects it, arguing:

If Aquinas were arguing for the conclusion that only an actual thing can cause another thing to actualize some property, then there would be no reason to believe that an actual blue thing cannot cause itself to actualize the property of being green. The blue thing is an actual thing, after all. But such self-causation is precisely what Aquinas is anxious to deny.

This is arguably not the self-causation that Aquinas is anxious to deny. Either this is supposed to be per accidens (thing that is blue, but not insofar as it is blue, causing itself to become a thing that is green) or per se (actual blue as such causing itself to become green). Aquinas follows Aristotle in dividing the two from each other, and per accidens has a different (and entirely derivative) account. As for per se, Aquinas elsewhere argues that something can actualize its own potential in the sense that one part can provide the act to the potential of another part, and natural changes can be actualized by the natures of things that undergo them in such a way that the generating cause of the nature is the mover. In other words, there are two things in the vicinity of this that are definitely not counterexamples to what Aquinas is saying: if the blue changed itself to green in such a way that part of the blue changes another part to green, this would not be the relevant kind of self-causation; and if the blue is changing to green because it is in its nature to do so, the generating cause of the blue, which gives it that nature involving that change, is not the relevant kind of self-causation. These are loose kinds of self-causation. The self-causation Aquinas needs to rule out is strict self-causation, as it would be defined by his interpretation of the Aristotelian account of change.

And the argument that Aquinas gives for ruling it out is that "it is not possible for the same thing to be at once actual and potential inasmuch as it is the same, but only inasmuch as it is different." Admittedly, this is be somewhat obscure on its own, but the claim has a parallel passage in the Summa Contra Gentiles:

Nihil idem est simul actu et potentia respectu eiusdem. Sed omne quod movetur, inquantum huiusmodi, est in potentia: quia motus est actus existentis in potentia secundum quod huiusmodi. Omne autem quod movet est in actu, inquantum huiusmodi: quia nihil agit nisi secundum quod est in actu. Ergo nihil est respectu eiusdem motus movens et motum. Et sic nihil movet seipsum.

The hujusmodis are a bit of pain for translation, but I suppose roughly: "Nothing that is the same can be at once actual and potential with respect to the same thing. But all that is changed, as such is potential: because change is the act of the existent inasmuch as it is potential. But all that changes is actual as such: for nothing acts except according as it is actual. Therefore nothing with respect to the same change is both moving and moved. And so nothing moves itself."

Recall that in an Aristotelian account of change, change is incomplete act, but most importantly it is the act of the changer in the potentiality of the changed. There is not an act of changing and then an act of being changed; it's just one act. So the point is that if something that is changed is so because its act of changing itself is its act of being changed, then it is both merely potential (not having G but capable of it) and also actual (being made to have G) in the same way (namely, according to the act that is the change) at the same time. That's a contradiction.

Now, you could perhaps be very stubborn about this and say that nonetheless the Latin can be read as saying something closer to what Almeida means. And probably that's not wrong. But people have pointed out before that the scholastics had concepts like being G eminently. Thus the sun generates a man; but the sun was held to have more universal properties that are, so to speak, being drawn on in this particular case insofar as the man has less universal properties related to those more universal properties. And while I don't think it's necessary in the context of the ST (where I think he is giving a much more general version of the argument than he is elsewhere), Aquinas says things that sound very like it in lecture 10 of his commentary on Book VIII of the Physics, where he gives the argument for univocal causation first and then says that a similar argument can be run for equivocal causation, if you take 'same' in a less strict way. But even interpreting it thus, it seems to me that "If y is actually F and potentially G, and x is the cause of y actualizing G, then x must be G" is an inaccurate characterization, because according to usual formalization conventions, the G would have to be univocal, whereas this interpretation is precisely not assuming that what is the same between x and y is univocal.

Almeida doesn't place a vast amount of emphasis on it, but what often seems to give analytic philosophers particular trouble is Aquinas's fire example. There are several things that are going on in the misreading of it, but I think a great deal of it is not understanding how medieval examples work. Medieval scholastics don't generally give examples as illustrations the way we do; illustration is a rhetorical, and not a dialectical, use of examples, and would generally be regarded as out of place in a rigorous argument. Their understanding of an example, which derives from Aristotle, is that an example is an abbreviated induction, just like an enthymeme is an abbreviated deduction. And for the medieval scholastics, an induction was an argument in which you divided the possibilities and then either established or ruled out that something applies to each part of the division. Thus an example takes one kind of case, shows you that it gets the result, and leaves it up to you to finish the induction. Since the point is not to illustrate the conclusion or even to give you a paradigmatic case of it, but to give you an abbreviated form of an inductive argument, the standard practice is to pick a really obvious case without complications, because arguments are supposed to start with the more obvious rather than the less obvious. This is why medieval examples are so often extraordinarily trite, and why you keep getting the same ones over and over again -- they are usually taking cases that were commonly recognizable as very obvious, at least when such cases were available. Thus there is no implication, from Aquinas's giving the fire example, that every case is like the fire example. It's just that univocal causation is a more obvious case than equivocal causation, so he gives a univocal case.

I'm still reading Almeida's book, but so far it is quite interesting.