05 / 06

Trinity Monotheism Once More: A Response to Daniel Howard-Snyder

William Lane Craig


A response to Howard-Snyder's criticism of the Social Trinitarian model laid out in Philosophical Foundations of a Christian Worldview. Exploiting insights from mereology, I offer a possible alternative to my previously defended view that the Trinity is strictly identical to God, taking the Trinity to be a group concept instead of the name of a substance.


Daniel Howard-Snyder does not like Trinity Monotheism. I mean, he really does not like Trinity Monotheism! This much, at least, is demonstrated by his pointed critique of our proposed model of the Trinity in our Philosophical Foundations for a Christian Worldview[1]  In this response I hope both to profit from his critique and to show that his objections to our proposal are not near as devastating as he seems to think. Since a blow by blow commentary would be inordinately long, I shall focus on his principal objections found in the critical portions of his paper, sections 3 and 4.

The strength of our proposal lies in the fact that it does not rest content with a merely formulaic understanding of the Trinity. Rather we try to offer a model which actually shows how the Father, Son, and Holy Spirit can be three persons in one substance. Here is the model: God is an immaterial substance or soul endowed with three sets of cognitive faculties each of which is sufficient for personhood, so that God has three centers of self-consciousness, intentionality, and will. One of the weaknesses of Howard-Snyder’s critique is that he fusses terribly over the analogies and suggestions which we offer as a springboard for arriving at this model while having comparatively little to say about the coherence or acceptability of the final model itself. For example, whether the persons mentioned in the model ought to be characterized as parts of God is really quite incidental to the proposal and may be left to mereologists to decide. The issue of parthood arose only as a suggested way of explaining why the persons are divine. But that they are divine on the final model seems obvious, since the model describes a God who is tri-personal. The persons are the minds of God.

The model itself is disarmingly simple and prima facie coherent. Why could not a soul be so richly endowed as to possess three sets of cognitive faculties sufficient for personhood, even as our souls possess one such set? I see no apparent reason it could not, nor does Howard-Snyder offer one. So the question remaining will be whether this proposal is theologically unacceptable even if coherent. Here Howard-Snyder does have misgivings, which he raises in his section 4. So let us consider them first. If they are not too serious, then the questions about parthood and composition raised in section 3 will be philosophically interesting but not vital to the model’s success.

Theological Objections (Section 4)

In section 4.1 Howard-Snyder criticizes, not the model, but the principal analogy used as a springboard to arrive at the model, along with a remark made near the end of our exposition. Howard-Snyder’s first objection to our use of the mythological three-headed dog Cerberus as a way of getting at the question of how three persons can together be one being is that Cerberus is not one dog but “three partially overlapping dogs.” [2] I find this claim astonishing. Multi-headed animals exist not merely in mythology but in the real world (see Fig. 1).

Picture of a two headed turtle. Picture of a two headed snake.
(a) (b)
Fig 1: Some two-headed reptiles

In Fig. 1(a) we see a two-headed turtle. Or so it appears. If Howard-Snyder is right, we are actually viewing two partially overlapping turtles. Indeed, on Howard-Snyder’s view what we see in Fig. 1(a) may not even be a thing and is in any case not a turtle at all. This is contrary to common sense and the nearly universal opinion that Fig. 1(a) is a picture of a two-headed turtle. The metaphysician who wants us to believe that what we actually see here is something far more bizarre and recherché than a mutant turtle, namely, two turtles which overlap except for their protruding heads, had better have some pretty compelling arguments for thinking that this is the case. But Howard-Snyder offers none. He simply asserts it.

He does respond to the sensible objection that in the case of Cerberus there exists an unusual, three-headed dog having three, distinct canine “persons.” He responds, first, by claiming that since brains individuate mammals, there cannot be a single dog having three distinct, complete, individually functioning brains. Mere assertion! Just as there are two-headed reptiles, there are also mutant mammals which have two heads and so two brains (see Fig. 2).

Sepia photo of a two headed calf.

Fig. 2: A two-headed calf.

These creatures are typically dysfunctional, but there is no inherent reason why this must be so. We can imagine a fully functioning two-headed dog. In the case of Cerberus we may suppose that he is a normal member of his species and, when mating, sires fully functional three-headed offspring. Having three heads is built into the genome of this species, and only mutants would be unfortunate enough to lack three fully developed heads. Why should we agree with Howard-Snyder that such a thing is impossible?

His second response trades on the ambiguities familiar in debates over personal identity. Suppose Rover, Bowser, and Spike were surgically divided and supplemented with body parts so that three separate dogs exist after completion of the operations. It would seem that we now have three new dogs. But what happened to Rover, Bowser, and Spike? They still exist but cannot be identical to the three new dogs, since none of them was a dog prior to the surgery. In order to avoid the conundrums that this would occasion, Howard-Snyder thinks that we should say that each of Rover, Bowser, and Spike was a dog prior to and after the surgery.

Cases of this sort typically involve such conflicting intuitions that no solution stands out as clearly correct, and those that are offered all seem counter-intuitive. I have far more confidence that I know a two-headed animal when I see one than I do in the truth of any particular solution to the puzzles posed by such thought experiments. Hence, objections to the existence of a three-headed dog like Cerberus based on such puzzles are not at all compelling objections. Indeed, Howard-Snyder will find himself confronted with puzzles parallel to those that he raises. For there are a variety of canine species. Suppose that Cerberus looks something like a three-headed wolf. On Howard-Snyder’s view Cerberus (if there is such a thing at all) is despite appearances not a wolf but a pack of three wolves. Now suppose that Bowser is reconstructed by the surgeons using collie parts and that poor Spike, once a vicious guard dog, wakes up from the surgery to find that his brain has been incorporated into the body of a chihuahua. It would seem that a new collie and a new Chihuahua have begun to exist, or at least new collie and chihuahua hybrids, even though Bowser and Spike survive the operation. But then what were they before the operation? Was Bowser always a collie and Spike a chihuahua? Surely not! So are they now really wolves, which look like a collie and a chihuahua? Lest Howard-Snyder is tempted to bite the bullet and say yes, let us suppose that the surgeon lacked the canine parts to complete the operations and so incorporated Spike’s brain into the body of a cat or an alligator. Is Spike with his reptilian body still a dog, a cold-blooded, egg-laying wolf? That seems absurd.

I think we should want to say that after the operation new creatures do begin to exist with which Bowser and Spike are not identical, even though Bowser and Spike survive the operations. Perhaps Rover, Bowser, and Spike are three parts of Cerberus, say, his brains, which get incorporated into new animals. Or perhaps Rover, Bowser, and Spike are immaterial persons who become somehow connected with new bodies. Affirming that Rover, Bowser, and Spike survive the surgeries thus does not commit us to the view that they were three distinct dogs prior to the operations. After the surgeries we have three distinct animals, at most one of which is identical with Cerberus, the other two being new animals. Rover, Bowser, and Spike endure as either material or immaterial parts of those animals. Hence, Howard-Snyder has not proved that Cerberus was not, as the story says, a three-headed dog but three distinct, if overlapping, dogs.

Howard-Snyder’s second objection to the Cerberus analogy is that it is obscure. He complains that we do not have the foggiest idea what it means to say that Cerberus “supports” Rover, Bowser, and Spike.

This criticism of the analogy is without merit, since the relation in question appears nowhere in the analogy or in our discussion thereof but is imported by Howard-Snyder himself from a line later in our chapter. If someone were to come across a damaged copy of Philosophical Foundations in which the offending line and everything thereafter were lost, he would still have our full discussion of the Cerberus analogy and the exposition of our final model. He would lack nothing requisite for understanding the analogy or our proposed model of the Trinity suggested by it. To clear up all ambiguity: just as Cerberus is a single dog with three consciousnesses, so God is a single spiritual substance or soul with three self-consciousnesses.

Having dismissed the Cerberus analogy, Howard-Snyder then takes issue with our afore-mentioned summary remark following the explanation of the model: “God would therefore be one being which supports three persons, just as our individual beings support one person.” He makes very heavy weather about our use of the word “support,” which appears only here in the entire chapter, as though it were the linchpin of the model. In point of fact, the sentence was intended only to summarize what had already been explained, meaning no more and no less than what had been said before, and can be deleted from the chapter with no loss of content. For the record what I was trying to express was the idea that a soul’s cognitive equipment is explanatorily prior to the number of persons there are: the reason there are three persons in the Godhead is because God is endowed with three sets of rational faculties sufficient for personhood, just as in my case there is one person because my soul is equipped with a single set of rational faculties. We could perhaps say that personhood supervenes on rational faculties and in that sense depends on what sort of soul is involved.

So the Cerberus story is a suggestive and helpful analogy, and the language of “supports” is inessential to understanding the model.

In section 4.2 Howard-Snyder turns to a discussion of the model itself. He correctly observes that on the model God “is not a person,” though he cashes this out tendentiously as God’s lacking the cognitive equipment sufficient for being “a self-reflective agent capable of self-determination.” [3] This is very misleading, as though God were not on our view a personal being. But in fact on our view God has the cognitive equipment sufficient for personhood three times over and so is tri-personal. Thus, there is nothing objectionable in the implication of our view that God is not a person. That is part and parcel of Trinitarian orthodoxy. Unfortunately, Howard Snyder evinces a disturbing proclivity toward Unitarianism. Thus, the observation by an anonymous referee that “every model of the Trinity must say that God is not a person” draws this surprising retort: “So far as I can see, this assertion does little more than express the referee’s myopic view of the options, not to mention his or her disregard for the plain sense of Scripture and tradition.” [4] Here I am afraid that Howard-Snyder is seriously mistaken. The Church has uniformly rejected the claim of Modalists and other Unitarians that God is a person. Not even anti-social Trinitarians would be so reckless as to affirm that God is one person. We all affirm that God is personal, but Trinitarians reject the claim that God is a person.

Howard-Snyder draws out three consequences of the denial that God is a person which he finds objectionable. First, the opening verse of the Bible “In the beginning God created . . . .” is false. This objection is fatuous, since the Old Testament scriptures do not distinguish the persons of the Trinity. When they are finally distinguished in the New Testament, we find that God the Father is typically described as the Creator, with Christ as His intermediary.

Second, human beings cannot be made in God’s image, since God is not a person. Again, God’s not being a person does not imply that God “is void of all personal attributes.” [5] We are made in God’s image because we are endowed with rational faculties sufficient for personhood even as God is so endowed, the only difference being that whereas we each have one such set of faculties, God has three. But we are personal beings just as God is a personal being.

Third, the view implies an abysmally low view of the divine nature, since the Trinity exemplifies the divine nature and yet fails to be a person. “If God is not a person or agent, then God does not know anything, cannot act, cannot choose, cannot be morally good, cannot be worthy of worship.” [6] The objection is misguided. Howard-Snyder assumes that God cannot have such properties unless He is a person. But it seems to me that God can have them if God is a soul possessing the rational faculties sufficient for personhood. If God were a soul endowed with a single set of rational faculties, then He could do all these things. By being a more richly endowed soul, is God thereby somehow incapacitated? How can augmenting God’s cognitive faculties make God less knowing, less good, less powerful, less worthy of worship? On our view it belongs to the divine nature to be a personal being, and a soul so richly endowed with rational faculties is, if anything, more majestic and worthy of worship than a more meagerly endowed soul.

In the final section of his paper 4.3, Howard-Snyder alleges that our view is not monotheistic. He makes the point that monotheism should not be defined so narrowly as to preclude Jews from being classed as monotheists. Jews and Christians agree that there is only one instance of the divine nature. But if we define the divine nature so as to include the property of being triune, we exclude Jews (not to mention Muslims) as monotheists. On the other hand, monotheists are united in the conviction that the divine nature includes the property of being worthy of worship, which implies being a person. According to monotheism something can be a God without exemplifying a nature that includes the property of being triune and nothing can be a God without exemplifying a nature that includes the property of being a person. On either score our view fails to be monotheistic.

This objection seems wrongly to assume that the word “God” must be defined in terms of the generic divine nature in such a way that all parties who can be said to be monotheists concur on the properties of that nature. There can be dictionary definitions of the word which are sufficient for ordinary usage without listing all the properties belong to the generic divine nature. So monotheists do not necessarily concur that the generic divine nature is not triune. But let that pass. Suppose we agree that the generic divine nature does not include being triune. Then the real sticking point is Howard-Snyder’s claim that all monotheists concur that God is a person. Here again, he confuses unitarianism with monotheism. Monotheists concur that God is personal, not that He is a person. Monotheists agree that one need not be a unitarian to be a monotheist. Thus, by Howard-Snyder’s reasoning, it is he who fails to be a monotheist.

In summary, then, it is noteworthy that Howard-Snyder’s objections to our actual model turn out to be quite weak.

Mereological Objections (Section 3)

We now come to what Howard-Snyder calls the “Diminished Divinity Problem,” in his section 3. This issue arises as a result of Leftow’s charge that if the persons of the Trinity do not instantiate the divine nature but are nonetheless said to be divine, then they enjoy at best a sort of diminished divinity. We replied that Leftow’s objection presupposed that there is but one way to be divine, which begs the question against Trinity Monotheism. That reply serves to defeat the objection; but in search of a positive account we explore the question, “In virtue of what are the persons of the Trinity divine?” We consider the analogy of felinity. There is more than one way to be feline, for a cat’s DNA or skeleton is fully and unambiguously feline, even though neither is an instance of the feline nature. On this score Howard-Snyder seems to agree: there is more than one way to be feline. Two questions then remain: (1) Why can a cat’s skeleton or DNA be truly said to be feline, and (2) Can the divinity of the persons of the Trinity be analogously explained? Notice that even if our answers to these questions are incorrect, that does nothing either to undermine the final model or to show that there is only one way to be divine. Leftow’s objection would remain defeated.

So, to consider the first question, why is a cat’s skeleton properly called a feline skeleton? We suggested the common sense answer that it is because it is part of a cat. It is a mere cavil when Howard-Snyder objects that transplanted organs from another class of animal can be part of a cat, for we were obviously talking about natural parts. Nor is it a serious problem when he objects that the atoms and molecules which are parts of a cat are not feline, for we can say that we are speaking of parts which are distinctive of the cat family, parts which a qualified biologist or paleontologist would be able to identify as belonging to a cat were he to be presented with them. This is just to say that such parts are unique to cats.

Rather Howard-Snyder’s principal objection to our analysis concerns the meaning of the words “there are two ways of being feline.” He then provides two—actually four, if one includes his comments in the footnotes—alternative ways of understanding what is meant by these words. Now it must be immediately said that what Howard-Snyder offers is not at all the meaning of these words but rather competing metaphysical analyses in terms of properties, exemplification, predicates, truth-makers, proper and improper parts, and so on. I am not being pedantic when I protest that none of these has anything at all to do with the meaning of the words in question, which are so simple that a junior high school student could understand them. They mean something like “There is more than one way of being cat-ish or cat-like.” The fact that Howard-Snyder’s analyses do not give the meaning of these words is evident from the fact that one could rationally assent to our sentence while dissenting from Howard-Snyder’s sentences. This is important because the Trinity Monotheist may not have any settled views on a metaphysical analysis while holding that the sentence is meaningful and true and that he can explain why it is true. If I were called upon to give an analysis of the claim in question, I should commit myself metaphysically to no more than

F. “Being feline” is truly predicable of any entity x if x either is a cat or is a distinctive part of a cat.

I am simply not prepared to take on all the metaphysical baggage that Howard-Snyder’s analyses would foist upon us.

For the sake of argument let us look at Howard-Snyder’s analyses in order to see their consequences for the two questions before us. The first analysis is in terms of a single property felinity being exemplified in two different ways. Howard-Snyder does not deny that such an analysis gives an adequate answer to question (1), but he denies that it will permit an affirmative answer to question (2). So let us have it aside for now and consider his second analysis. According to this analysis, “x is feline” may be used to attribute either of two properties. I note that Howard-Snyder’s introduction of linguistic expressions in this analysis is unmotivated. As long as one is going to use properties, why not say x has the property of being feline if it has either of the two further properties? Such an account would be in line with our claim that there is more than one way to be feline (that is, to have the property of being feline). But Howard-Snyder’s analysis splits felinity itself into two distinct properties, which seems in contradiction with our claim that there is more than one way to be (fully) feline (or divine). In any case Howard-Snyder objects to there being a second property of felinity. For the truth-maker for

3. Howard-Snyder’s cat Socrates’ skeleton is part of a cat.

is also the truth-maker for

2. Socrates’ skeleton is feline.

But on Howard-Snyder’s second analysis, the truth-maker for (2) ought to be a different, more complex fact. Therefore, (2) does not mean what the second analysis says that it means.

This objection to the second analysis gratuitously assumes that there are such things as truth-makers. One wonders how they slipped into our ontology unnoticed! The Trinity Monotheist is not obliged to align himself with that minority of philosophers who believe in truth-makers. Moreover, if there are truth-makers, why think that they are “facts” in this case (it has been complained that such a move is ontology by theft rather than by honest toil). I should say that the truth-maker of (2) and (3) as well as

1. Socrates is feline.

is just Socrates himself, the real, live cat. The second analysis in no way requires us to identify truth-makers as facts. Moreover, Howard-Snyder seems to assume that truth-making is closed under logical implication, so that if (3) implies (2), what makes (3) true also makes (2) true. But that assumption is false. For example, “Socrates has retractable claws” implies that “Grass is green,” since both are true, but they obviously have different truth-makers. Perhaps he thinks that (2) and (3) are synonymous. But that is clearly wrong, since something can be feline without being part of a cat. Thus, (2) can be false (Socrates’ skeleton is not a cat), even though (3) is true. Finally, even if the truth-maker of (3) did serve to make (2) true, that is inconsequential, since propositions can have multiple truth-makers. So the elaborate objection to the second analysis (misguided as it is) is unconvincing.

The third analysis is Frances Howard-Snyder’s ingenious suggestion that something can exemplify the property of felinity by being either a proper part of a cat or an improper part of a cat. What this third analysis really amounts to (and what distinguishes it from the first) is the claim that there is only one way of exemplifying the property of felinity: by being part of a cat. But there are both proper and improper parts of a cat. We husbands should listen to our wives, and in this case Dan would have done well to heed Frances’ suggestion, for if the only difficulties it inherits are, as he says, those attending the second analysis, then it is quite unobjectionable and perspicuous. Indeed, it is the most simple analysis of all: one property and one way of exemplifying it. Very nice!

Finally, what about Dale Tuggy’s fourth analysis, according to which the only way of being feline is to exemplify the nature of a cat, but “feline” is predicable of items that are not feline? If by “predicable” Tuggy means “truly predicable,” then on pain of incoherence, he must mean that parts of cats can be truly said to be feline without their exemplifying the nature of a cat. [7] That is correct, but if he is prepared to admit properties such as a cat nature, it is unclear why he thinks there is no property corresponding to the predicate “feline” when ascribed to a cat’s skeleton. Still, his analysis is helpful in pointing us to predicates rather than properties and so to a more metaphysically economical analysis such as (F) above.

Now if Howard-Snyder has failed in his attempt to undermine the adequacy of the proffered analyses of the claim that there are two ways to be feline, what about the applicability of the first analysis to the problem of the Trinity? According to that analysis there will be one property of divinity which both the Trinity and the persons exemplify. Howard-Snyder rightly observes that at least four items, then, have this property. But in that case, he alleges, the rationale we offer for thinking that there is exactly one God vanishes.

True, the Trinity ‘as a whole’ exemplifies the property of being composed of the Persons while no Person exemplifies it. But . . . the divine nature cannot include that property since there is one and only one complex property whereby a thing can be divine and each of the Persons exemplifies that property but each of them lacks the property of being composed of the Persons. So . . . there are four items that exemplify the divine nature and hence there are four Gods. If, however, . . . the divine nature does include the property of being composed of the Persons, then none of the Persons is divine since . . . there is no other nature whereby a thing can be divine. [8]

This reprise of Leftow’s objection is as question-begging as the original. It assumes that the only way of being divine is to exemplify the divine nature. This is precisely what Trinity Monotheism denies. The property of being divine, like the property of being feline, is a single property which a thing may have as a consequence of various factors. One reason something is divine is because it exemplifies God’s nature. Another reason, we suggest, is that it is a distinctive part of God. If, then, the persons of the Trinity are distinctive parts of the thing that instantiates God’s nature, they are divine, and the analogy of divinity with felinity is tight. Thus, on the analogy with (F) we may affirm

G. “Being divine” is truly predicable of any entity x if x either is a God or is a distinctive part of a God.

The question then remains whether on our model the persons should be considered to be distinctive parts of God.

In section 3.2 of his paper, Howard-Snyder presses his critique forward based upon his second analysis, according to which there are two distinct properties ascribed by the ambiguous predicate “is divine.” Since we reject that analysis, this section of his paper is irrelevant to our model. Still, since interesting questions arise in this section, let us pursue the discussion. Here we encounter the very difficult question of whether the persons of the Trinity are parts of God. I think our final model leaves this an open question. Whether or not they qualify on our model as parts, I think the persons on our model are indisputably divine, for they are God’s persons and are omnipotent, omniscient, omnipresent, worthy of worship, and so forth. Neither is their being parts crucial to their not being three Gods, for they are clearly not three separate beings on the model, there being only one soul endowed with three sets of cognitive faculties. The model is what it is regardless of how we decide the mereological question.

In his fine book Parts Peter Simons catalogues the various possibilities of composition, as that notion is commonly understood (Fig. 3). [9] He analyzes composites in terms of individuals, masses, and collections. An individual is anything which can be the subject of a true singular count predication which is neither disguisedly plural nor disguisedly mass, for example, “a man.” Collections are pluralities, that is to say, objects which are essentially not one thing but many things, such as groups and classes. Masses are concrete particulars which are neither one nor many individuals. When Simons uses “some” with a mass term, as in “some water” he writes “sm” to differentiate this usage from “some” in the sense of part, as in “some of the water.”

Fig. 3: Possibilities of Composition

________ can be composed/made of   ________  
Category Category Example
individual individuals wall/stones
individual mass sweater/sm wool
individual individuals + mass toffee apple/an apple+sm toffee
individual individuals + mass fruitcake/currants+sm dough
mass individuals gold/gold atoms
mass masses dough/flour+water
mass mass(es)+individuals blood/plasma+blood cells
collection individuals pack/wolves
collection mass snowballs/sm snow
collection individuals+mass toffee apples/apples+sm toffee

Which of these possibilities might seem appropriate for the Trinity?

We may rule out immediately any composites involving masses, since neither the Trinity nor the persons can be construed as masses. That leaves an individual composed of individuals and a collection composed of individuals as the two candidates. Are either of these appropriate?

Consider a collection made up of individuals. The Trinity has obvious affinities to a group. The Trinity seems to be a plurality, not an individual, which is made up of the three divine persons. We naturally speak of the persons as members of the Trinity, which is the language associated with groups and classes. A group is composed of several individuals fulfilling certain constitution conditions. The roles played by the persons in the economic Trinity would be such a condition constitutive of a group. The Trinity would be an unusual group in that it has its members essentially, but it would seem to be a group nonetheless.

Regarding the Trinity as a group would easily dissolve Howard-Snyder’s objection raised in section 3.2 that since parthood is transitive the parts of the Father would be parts of the Trinity, for parthood is not transitive across types of composition: John is part of the team, but John’s nose is not part of the team. Similarly, the Father’s faculties are not members of the Trinity. As members of the Trinity, which is composed of divine persons, the Father, Son, and Holy Spirit are uniquely worthy of worship. As for Howard-Snyder’s complaint that what makes the persons divine is not their being part of the Trinity but their possessing the properties of divinity, one could either just agree with him on that score or one could insist that their being members of the Trinity is explanatorily prior to their possessing such properties. After all, the Trinity is not a group that one joins because one has the requisite properties; rather the members have the properties precisely because they are members of the Godhead.

If we do regard the Trinity as a group, then the claim of Trinity Monotheism that the Trinity is identical with God will have to be given up. For God is clearly an individual substance and, hence, an individual. The Trinity would not be some substance in addition to God; rather it is simply the collection of the divine persons, which is not itself an individual or a substance. Our final model of God as an immaterial substance enjoying three centers of self-consciousness would remain unaffected. One would merely not identify the Trinity with that substance, nor would it follow that the persons are parts of God in virtue of being parts of the Trinity.

One drawback of this view is that the Trinity then involves no more unity than is involved in group notions. The persons of the Trinity should be not just a triad but a triad in unity, a Trinity. One seems in danger of lapsing back into Swinburne’s tri-theistic model. But it might be rejoined that the unity is preserved by our model rather than by the group notion itself. There is only one God because there is only one immaterial substance having three sets of rational faculties each sufficient for personhood. The unity is grounded in the model’s concept of God, whereas the Trinity is a group notion which we form by collecting the persons together. In view of the unity secured by the model, God is more like Cerberus than a pack of wolves. We can consider the trinity of canine “persons” Rover, Bowser, and Spike as a group without identifying them as parts of Cerberus or thinking that Cerberus is a pack of dogs rather than one dog. If we go this route, then our view should no longer be classified as Trinity Monotheism; rather we have enunciated a different view, uncontemplated by Leftow, which deserves a different name.

Suppose instead that we regard the Trinity not as a group but as an individual composed of individuals. Since that individual is naturally to be identified with God, we thereby stick with Trinity Monotheism. The Trinity just is the tri-personal spiritual substance described in our model. The individuals composing this being are the three persons of the Trinity. On this view Howard-Snyder’s transitivity objection also fails. First, since the Father’s faculties are not individuals, one is illicitly assuming once again transitivity across different types of composition. The Father is not an individual made up of individuals. Second, the Trinity Monotheist may in any case appeal to a special sense of parthood according to which being a part involves making a direct functional contribution to the whole. [10] Parthood is not in this case transitive. For example, a nucleus is a part of a cell but not a part of the body composed of cells (the body does not have a nucleus), and a handle is a part of the door of a house but not a part of the house (the house does not have a handle). The Trinity clearly has only three parts in this sense. (Such a sense of parthood may also be applicable in the case of parts of a cat which are distinctively feline.) Furthermore, the question of the divinity of the persons seems no more problematic on this view than on the view that takes the Trinity to be a group. So Howard-Snyder’s worry about what makes the persons divine is put aside.

The more difficult question which remains for this view is the question raised by Howard-Snyder in section 3.3: are the persons individual substances? The answer to that question will depend on whether one thinks that inseparable parts of a substance are themselves substances. The example of the wall composed of stones is not illuminating because the stones were substances before being made into the wall, and the wall can be separated into stones which would then be the same substances. What about parts which were never separate substances, such as one’s hands? Are they substances? What about parts which cannot be separated from the whole without undergoing substantial change, such as the trunk of a tree? Are they substances? Such parts are individuals, but it is not clear that they are substances. They seem to lack the “stand alone” quality that something must have in order to be a substance. If they are not substances, they may still have, however, enough integrity to have natures. A hand, for example, seems to have certain essential properties, such as having digits and having an opposable thumb. The persons of the Trinity could similarly share a certain nature, just as my hands do, without being substances in their own right. That nature would include all the great-making properties that make them worthy of worship. In such a case we should have three parts composing one substance, as in traditional Trinitarian formulas.

On the other hand, suppose we say that inseparable parts can count as substances in their own right. In that case the persons of the Trinity would doubtless count as individuals who are substances. They would again share the nature of a divine person. But never mind: as inseparable parts they are still three persons in one substance. They are no more instances of the nature of that unique substance than my hands are instances of the human nature. So the unity of God is preserved along with the divinity of the persons. Hence, nothing of significance hangs on whether we regard a substance’s inseparable parts as substances. The crucial fact is that these individuals compose one unique, indivisible individual which is a substance.


In conclusion, it seems that our Trinitarian model withstands Howard-Snyder’s criticism. The model is theologically unobjectionable and open to various mereological construals, leaving it up to the metaphysician to choose that construal which accords best with his views.

  • [1]

    William Lane Craig and J. P. Moreland, Philosophical Foundations for a Christian Worldview (Downer’s Grove, Ill.: Inter-Varsity, 2003), pp. 575-96.

  • [2]

    Daniel Howard-Snyder, “Trinity Monotheism,” Philosophia Christi 5 (2003): 396.

  • [3]

    Ibid., p. 399.

  • [4]

    Ibid., p. 400.

  • [5]


  • [6]

    Ibid., p. 401.

  • [7]

    Ibid., p. 389.

  • [8]

    Ibid., p. 387-8.

  • [9]

    Peter Simons, Parts (Oxford: Clarendon Press, 1987), p. 232.

  • [10]

    Ibid., pp. 107-8.