The question concerning the possibility of miracles is connected to a wider question concerning the relation of secondary or natural causes and the causa prima, God. How do these two 'orders' of causation fit together?
1. One extreme position is occasionalism according to which all causal power is exercised by God. For the occasionalist, God is all-powerful not just in the sense that he can do all that is (broadly) logically possible, but also in the sense that he exercises all the power that gets exercised. For the occasionalist, God is the only genuine cause and all secondary causes are mere occasions for the exercise of divine power. Although I have defended occasionalism elsewhere ("Concurrentism or Occasionalism?" Am Cath Phil Quart, Summer 1996, 339-359), I will not be assuming its truth here.
2. At the opposite extreme there is deism which is essentially the view that God is an initiating but not a sustaining cause of the existence of the universe. God is accordingly a mere cosmic 'starter-upper': he created the universe or the initial segment thereof in principio (taking this phrase temporally) but ever thereafter it has managed to exist in its own. On this view secondary causes are 'true' (productive) causes, and God in no way either sustains them or co-operates with them. One might speak colorfully of God's abdication after the cosmic start-up. Contrast this with occasionalism's God who is a sort of cosmic 'control freak.'
3. A middle position is concurrentism. When two or more causes unite to produce an effect what we have is a concursus. We speak of a concurrence of the concausae, the co-causes or cooperating causes. When two or more causes work together to produce an effect, no one of them is sufficient for the effect; they are sufficient only jointly. On concurrentism, then, God cooperates with natural causes to produce effects. When heating causes boiling, the heating, contra occasionalism, is causally efficacious, but God is co-efficacious.
Now Kant rejects concurrentism. God does not cooperate in the sustaining or conserving of substances for the reason that he is their sole ontological support. Nor does God cooperate in any alterational change in a substance. For when it comes to changes in a substance, natural causes suffice. Kant writes in his Lectures on Philosophical Theology, p. 147:
Applying this to God, it is clear first of all that he does not concur in the existence of substances. For substances contribute nothing to their own duration, and therefore they cannot themselves operate in union with God as concausae of their own conservation. In this case, there is only a subordination of causes, so that every substance has its ground in God as the prima causa, since the matter of every substance itself is created by him. But just for this reason, there can be no concursus. For if there were, the substance would have to be coordinated with God.
In the same way, there is no concursus of God in natural events. For just because they are events given in nature, it is presupposed already that their first proximate cause is in nature itself. Hence this cause must be sufficient to effect the event, even if the cause itself (like every natural cause) is grounded in God as the supreme cause.
4. Kant is surely neither an occasionalist nor a deist. And we have just seen him reject concurrentism. I submit that Kant's position is usefully labelled 'conservationism.' It is a fourth position and a second moderate position. Accordingly, the sole role of the causa prima is the conservation of substances; he has no role to play in the causation of their changes of state. All event-causation within nature is wholly explainable in terms of other event-causes in nature. Without God, there would be no nature and hence no event-causation in nature; but God is not involved in natural changes. It seems to follow that in Kant's scheme there is no conceptual space for miracles ontically construed. Kant writes, pp. 151-152:
. . . from what we have already said about God's cooperation with natural events, it can be seen how inappropriate it is to use this expression in place of divine conservation. For substances are certainly not coordinated with God, since they wholly depend on him as their causa solitaria absoluta prima. So how can I regard substances as concausae, concurring with God in their own duration? Would I then not be asserting that their existence is not actuated by God, and that they do not have need of him alone for their duration, since he is only a cooperating cause of it?
It is equally wrong to posit a concursum Dei for events given in nature. For we can always think of a causa proxima for these events, operating in accordance with laws of nature; since otherwise they would eo ipso not be events given in nature. So it is likewise unthinkable that God, who is the causa prima of the whole of nature, might also operate as a concausa in each particular event. For then these events would only be so many miracles. Every case where God himself acts immediately is an exception to the rule of nature. But if God is to cooperate as a special concausa of every particular event given in nature, then every event would be an exception to the laws of nature. Or rather there would be no order at all in nature, because events in it would not happen according to universal rules, but in each God would have to give a complementum ad sufficientiam to anything which was to be set up in accordance with his will. But we could not think of such an imperfect world united to a wise author.
5. Miracles, then, are complementa ad sufficientiam. They are additions from outside the secondary causal nexus that complement a secondary causal input and bring it up to the level of causal sufficiency. In the passage just quoted, Kant seems to be saying that miracles ontically construed are "unthinkable" and thus impossible. For every natural event has a proximate sufficient cause in other natural events. But that amounts to saying that the laws plus the past necessitate the events of the present, so that there is no possibility of any divine interference.
To put it another way, if any natural event had an insufficient secondary cause, and so needed complementation from the Primary Cause, then nature would not be the orderly, law-governed system that it is. A system of nature that needed to be helped along by divine interference would be defective and not what one would expect from a wise author.
6. Although I have just quoted Kant to the effect that miracles, ontically construed, are "unthinkable," a few pages earlier, on p. 147, he says the opposite: "For it is always thinkable that a natural cause is not by itself sufficient to produce a certain effect. In such a case, God might give it a complementum ad sufficientiam. . . ." Whether this contradiction in Kant's texts is real or apparent, it testifies to the difficulty of finding conceptual space for miracles.