Building on the substance dualist view that our own individual substance/soul supports one mind (i.e., person), Williams, Craig, and Hasker propose models of the Trinity according to which a single trope of divinity—a divine substance/soul—supports three minds (i.e., persons) in like manner. Craig relies on this claim to flesh out how three persons can be said to compose one being, and Hasker relies on the claim to flesh out how three persons can be said to be constituted by one being.
Daniel Howard-Snyder will have none of this. In response to Craig, he writes:
Craig’s use of “supports” has no precedence in the English language. Suppose there is some x, y, and z such that x supports y and z. Why should we infer that, therefore, y and z compose x? The foundation of my house supports its walls, floors, roof, and so on, but they do not compose the foundation. The worry here is intensified by the fact that there is no use of “support” and “compose” in ordinary parlance such that supporting entails composing, as a look at the Oxford English Dictionary will reveal. Without, at minimum, a stipulative definition in terms that we can understand, we have no idea what [Craig means] by “supports” and hence we have no idea what [Craig means] by the claim that the Persons compose the Trinity “as a whole” because the latter supports the former; we have no idea what proposition is expressed, we have no idea what model is proposed for our consideration. (Howard-Snyder, “Trinity Monotheism,” pp. 120-121).
And similarly in his review of Hasker:
[F]ew of [Hasker’s] peers, if any, will understand what he means by “support”. At some level, he is aware of the problem since he seeks “a more precise account of the relationship between the persons and the divine nature than is provided by the loosely defined ‘support’ relation” (237). However, the problem is more acute than Hasker acknowledges since nowhere does he define it. He only says that “the term is used in the ordinary sense in which we can say that the human body/mind/soul…‘supports’ the continuing conscious life of a human being” (228). This is no definition, not even a “loose” one; nor is there any such thing as “the ordinary sense” of the term “supports” that is used to say that “the human body /mind/soul supports the continuing conscious life of a human being”. Thus, the primary explanatory relation posited by the first model is an explanatory surd. (review of Hasker, Metaphysics)
Is the “supports” claim Craig and Hasker make really as unintelligible as Howard-Snyder thinks? I think not. A fairly straightforward case can be made that the relevant “supports” relation here should be understood in terms of parts being grounded in—i.e., supported by—a whole, where a ground is a relation of non-causal ontological dependence. Here are three examples where this makes perfect sense:
1. The Mountain and the Peak
The peak of a mountain ontologically depends on and is posterior to the mountain as a whole. Yet the peak of a mountain is undeniably a part of the mountain, and so (partially) composes the mountain. So understood, I see nothing untoward in saying “the mountain supports its peak.”
2. The Circle and the Semi-Circles
Draw a circle. Now draw a partitioning line creating two semi-circles. The two semi-circles ontologically depend on and are posterior to the circle as a whole. Yet the two semi-circles are undeniably parts of the circle, and so compose it. So understood, I see nothing untoward in saying “the circle supports its semi-circles.”
3. The Fabric and the Flag
A piece of fabric constitutes a flag in flag-favorable circumstances (e.g., certain arrangements of colors, patterns, and symbols). The flag ontologically depends on and is posterior to the fabric, yet the flag is constituted by the fabric. Arguably, we might even say the flag is a part (albeit an improper part) of the fabric. So understood, I see nothing untoward in saying “the fabric supports the flag.”
So it seems to me “supports” is a perfectly intelligible word to use to describe the relation between a whole and its parts, where the parts are ontologically dependent on and posterior to the whole. That this is in fact how Craig understands his model is made clear by the fact that he (with Moreland) says as much in their discussion of substances earlier in Philosophical Foundations (pp. 219-223). There they distinguish substances from “property-things” and argue that substances as wholes are ontologically prior to their parts, whereas property-things as wholes are ontologically posterior to their parts (I have a discussion of this distinction here). If Craig’s and Hasker’s Trinitarian models are problematic, it is not because their use of a “supports” relation cannot be given intelligible meaning.
The problem with the claim that a substance “supports” the three divine persons is that it is incompatible with the divine perfection of aseity. It seems clear that “supports,” for Craig and Hasker, tracks ontological priority, as it does in the three examples above. “Supports,” so conceived, is an asymmetric ontological dependence relation. But if the three divine persons asymmetrically depend for their existence on an underlying substance then they do not exist a se and so are not perfect and so are not divine.