The Problem of Triunity

There is a venerable logical problem about the doctrine of the Trinity that is often framed as a set of inconsistent claims:

1. (i) There is exactly one God.

2. (ii) There are exactly three divine persons.

3. (iii) Every divine person is God.Footnote ¹

Although each of these statements is a commitment of the orthodox doctrine of the Trinity, it appears that they cannot all be true. Virtually all of the literature on the Trinity in contemporary analytic theology is devoted to this traditional problem.Footnote ² But in the following, I argue that there is another logical problem about the Trinity that deserves attention – a problem which I will call The Problem of Triunity. Rather than proceeding from the fact that God is one and the persons are many, as the traditional problem does, the Problem of Triunity proceeds from the fact that, in one sense or another, God is many, and yet each divine person on his own is just one.

The Athanasian Creed declares that Christians worship ‘one God in Trinity and Trinity in Unity’ (Sullivan (Reference Sullivan1907) ) – a statement not unlike an even terser Trinitarian affirmation that is as familiar to Christian thinkers as it is prima facie unobjectionable:

1. (1a) God is triune.

Although this claim does not wear its meaning on its face, it is widely endorsed in one form or another. Thus, Augustine identifies God with the Trinity as a whole, saying that ‘the Trinity is the one and only and true God’ (De Trinitate, I.i.4), and a number of contemporary authors make similar claims, identifying God with a triune substance or property bearer (Craig & Moreland (Reference Craig and Moreland2003), Plantinga (Reference Plantinga, Feenstra and Plantinga1989), Wierenga (Reference Wierenga2004), and Yandell (Reference Yandell, McCall and Rea2009) ). Still other authors identify God with a substance that lives a triune life (Leftow (Reference Leftow2004) and Effingham (Reference Effingham2015) ), and this, too, seems to be a strategy for cashing out the claim that God is triune. So, although (as we shall see later) orthodox Christians might be able to deny that God is triune in a strict sense, I take it that (1a) is plausible and widely affirmed.

Can we say more about what it means for God to be triune? Whatever else it may mean, triunity minimally involves being in some sense three-personed. The exact sense of personhood relevant to the doctrine of the Trinity is a matter of dispute, but fortunately it is a dispute we can ignore for the time being. Just treat the claim that God is triune as a sort of disjunctive predication, with a disjunct for each of the leading accounts of the doctrine's notion of personhood.

Another claim that won't raise many eyebrows among orthodox Christians is the claim that each member of the Trinity on his own is not triune. For ease of exposition, I will take just one of the divine persons, the Son, as an example. So:

1. (2a) The Son is not triune.

On the face of it, this claim is eminently plausible. For, again, to say that God is triune is to say (at least) that God is three-personed. But the Son is just one person – one member of the three-membered Godhead.Footnote ³

And now trouble looms, for it seems to follow from (1a) and (2a) that the Son is not God. Yet, proposition (iii) in the traditional problem of the Trinity above – one of the commitments of the orthodox doctrine of the Trinity – entails that:

1. (3a) The Son is God.

Clearly something has gone wrong, for (1)–(3) form an inconsistent triad; any two of them entail the negation of the third, and so at least one of them must go. Which will it be?Footnote ⁴ I propose to consider each of them in turn, weighing the costs of various possible solutions. Let's take them in reverse order.

Denying (3a)

Can we deny (3a), the claim that the Son is God? Some writers – typically social Trinitarians – argue that we should read statements like (3a) as predications rather than identity claims (e.g. Craig & Moreland (Reference Craig and Moreland2003), Swinburne (Reference Swinburne1994), Wierenga (Reference Wierenga2004) ). Thus ‘The Son is God’ should not be understood as saying that the Son is identical to God, but rather that the Son is divine, which is not obviously inconsistent with (1a) and (2a). How exactly is the predicate ‘is divine’ satisfied? Using the property felinity as an example, Craig and Moreland argue that there are at least two ways to satisfy it. Something can be feline by being an instance of cat-nature, but a cat's skeleton or DNA are feline as well, and yet surely they do not instance the nature of a cat. They are merely distinctive parts of a cat. Analogously, Craig and Moreland claim that the Trinitarian persons can be divine either by instantiating the divine nature, or by virtue of being distinctive ‘parts’ of God (590).Footnote ⁵ This gives us two potential strategies for solving the Problem of Triunity. We read (3a) as the claim that the Son is divine, and then say that the Son satisfies the predicate either by instantiating the divine nature, or by being a distinctive ‘part’ of God. In either case it is not obvious that any conflict remains with (1a) and (2a).

Craig and Moreland adopt the latter approach, claiming that the Son is divine by virtue of being a distinctive part of God, while explicitly denying that the Son possesses the full divine nature. Though others will regard this move as radical, Craig and Moreland welcome the conclusion that the Son does not have the full divine nature, insisting that the Trinity alone, and not the individual persons, instantiates that nature, and that is why there is just one God (Craig & Moreland (Reference Craig and Moreland2003), 590). However, Craig and Moreland's position on the divinity of the persons does not seem to be popular. In general, the traditional doctrine of the Trinity is understood as affirming that the Son (as well as the other divine persons) has the full divine nature. The creeds are clear that the Son shares the very same nature as the Father, and while there is dispute about whether that shared nature is concrete or abstract, as the creeds are typically understood, that shared nature is no less than the divine nature – the full nature of God. Consequently, most Christian philosophical theologians will not accept this solution to the Problem of Triunity.

Turn to the other option. The Son is divine by virtue of possessing the divine nature – the set of properties individually necessary and jointly sufficient for divinity (omniscience, omnipotence, etc.). If we pursue this approach, we can read (3a) as:

1. (3b) The Son possesses the divine nature.

This appears to be consistent with (1a) and (2a). But in the end I think that it is just trading one problematic triad for another, because (1a) has a relevant cousin:

1. (1b) Triunity is a property included in the divine nature.

That is, triunity is a member of the set of properties both individually necessary and jointly sufficient for being divine. If there are good reasons to think that (1b) is true, then we have merely exchanged our original problematic triad for a new one:

1. (1b) Triunity is a property included in the divine nature.

2. (2a) The Son is not triune.

3. (3b) The Son possesses the divine nature.

In my experience, many Christian thinkers are reluctant to deny (1b), and it will be worth our time to try to tease out possible sources of this reluctance. Unfortunately, we will find that there is a respectable case to be made on (1b)’s behalf.

Craig and Moreland (Reference Craig and Moreland2003, 590) suggest that (1b) follows from God's being essentially triune. But why think that triunity must be included in the divine nature just because it is one of God's essential properties? The unstated premise seems to be that any property God has essentially is a property included in the divine nature. This, however, can be reasonably denied. Just as I can be essentially human and yet have essential properties that are not included in human nature (e.g. perhaps the world-indexed property being the author of this paper in W, where W is the actual world), so God can be essentially divine and yet have essential properties that are not included in the divine nature. So Craig and Moreland's argument for (1b) fails.

Nevertheless, (1b) boasts its share of attractions. Here are three better reasons to suppose that triunity is included in the set of properties individually necessary and jointly sufficient for being divine.

First, an inductive reason. Triunity has a lot in common with the standard divine attributes. Like those attributes, it is one of God's essential properties; like those attributes, it is a property for which God is worshipped; like those attributes, it intuitively plays some sort of kind-defining role – it tells us something about what kind of being God is. So it is at least very natural to suppose that triunity is a divine attribute. Relatedly, there is a certain tidiness, or theoretical elegance, to the idea that everything there is to say about God's essence and about what kind of being God is, including triunity, is summed up by God's being divine.

Second, a perfect being reason. It is sometimes said that triunity is a great-making property. Interpersonal relationships, loving and cooperative communities, and the like are highly valuable. Ceteris paribus, being in such relationships is better than being in isolation. (Just imagine yourself behind a veil of ignorance, presented with a choice between two lives. All you know about the two lives is that one will be a life of isolation, and the other a life of loving community. Which would you choose?) If loving communities are highly valuable – more so than states of isolation (ceteris paribus) – then it seems that a triune being that contained that sort of value within its own borders would be a greater, more valuable being than an otherwise similar unitarian being. In other words, triunity is a great-making property.Footnote ⁶

Objection: triunity is incompatible with other candidate great-making properties like being a person, being conscious, being a moral agent, etc., so on balance a triune being might not be greater than the greatest unitarian beings after all.Footnote ⁷ But properties like these have Trinitarian analogues; instead of being a person there is being three-personed; instead of being conscious there is (on some models anyway) being conscious three times over; and so on. These Trinitarian properties are not clearly less valuable than their unitarian counterparts, at least provided that there is no disharmony between the Trinitarian persons.

Another objection: triunity may not be a great-making property because it may not be a property at all, or at least not one that any being can instantiate. For a plurality (and a fortiori a plurality of three persons) does not have properties, and even if there is such a thing as the fusion of three persons, that fusion is not itself three persons; it merely has three parts that are persons.Footnote ⁸ This objection requires that we take a closer look at the nature of triunity. I have claimed that triunity involves being three-personed, and the objection points out that no individual thing is identical to three persons. But I do not think that the property of being three-personed needs to be construed as the property of being identical to three persons. For example, why wouldn't something composed of three persons be aptly described as three-personed? In fact, there are possible beings discussed in the Trinitarian literature that seem aptly described as three-personed, and yet are not identical to three persons. I offer two examples. First, Craig and Moreland (Reference Craig and Moreland2003, 591) suggest that there could be a soul which was endowed with three sets of cognitive faculties, each set being ‘sufficient for personhood’. So the persons, they claim, are parts or part-like constituents of the soul. It seems to me that such a soul is aptly described as three-personed. Second, Effingham (Reference Effingham2015) defends the possibility that a rational substance multilocates in such a way that there are different ‘versions’ of the substance (as one might speak of older and younger ‘versions’ of a backward time-traveller). He argues that, in certain cases, the different versions of a multilocated rational substance will be distinct persons. If there are three such persons, then once again it seems to me that the one multilocated substance is aptly characterized as three-personed. Arguably, then, it is possible to be three-personed.

So there is a strong prima facie case that triunity is a great-making property, and it is widely held (by so-called perfect being theologians) that to be divine is to be maximally great. If so, then perfect being theology recommends including triunity in the divine nature.

Third, an explanatory reason. Consider the question: why is God triune? (1b) allows for a simple answer: God is triune because God is divine. And there is reason to think that we don't have many other options here, for we might take the doctrine of divine aseity to entail that it is the very nature of deity to exist (be instantiated, or whatever). Then nothing (aside from deity itself) can explain/ground the fact that there is a divine being. It follows that triunity cannot explain this fact, nor can there be some third fact which explains both why there is a divine being and why that being is also triune. So unless deity explains triunity, then the fact that the divine being is also triune either is, or is explained by/grounded in, some independent brute necessity. Of course, one might just bite this bullet, but this is biting a bullet, because we prefer to explain things when we can, and, as Pruss (Reference Pruss, Moreland and Craig2009, 45) points out, philosophy is very much in the business of explaining necessary truths.Footnote ⁹

But prima facie, (1b) is not the only way to secure the desired explanatory connection between divinity and triunity. In particular, one might think that, instead of including or entailing being triune, perhaps the divine nature merely entails belonging to a trinity (cf. Swinburne (Reference Swinburne1994), ch. 8). So instead of (1b) we might venture:

1. (1c) Being a member of a trinity is included in the divine nature.

As far as I can see, this substitute is consistent with (2a) and (3b), for the Son can be a member of a trinity without being triune. However, there are reasons to be unhappy with (1c). For it seems obvious, trivial, even platitudinous that God – that is, the one and only God that Christians confess – is divine. But clearly God – the one God – is not a member of a trinity. Rather, God is triune, and only each individual divine person is a member of a trinity. So it looks very much like we have a counterexample to (1c).

How might a defender of (1c) reply to this objection? Maybe the one God can be counted as divine simply by virtue of ‘including’ persons that possess or instantiate the divine nature. Then one could deny that being divine in this ‘derivative’ way entails being a member of a trinity, even if being divine in the ‘direct’ way that the persons are does entail being a member of a trinity. But not everyone will be satisfied with this; there is certainly something odd about the suggestion that God is only derivatively divine.Footnote ¹⁰ Alternatively, a defender of (1c) might say that the one God is a member of the Trinity; indeed, God is (identical to) each member of the Trinity. I will consider this kind of proposal below when I examine ways to deny (1a), the claim that God is triune.

Another reason to be unhappy with (1c) is that, even if it secures a suitable explanatory connection between divinity and triunity, it does nothing to address the other reasons for thinking that triunity is a property included in the divine nature: that triunity itself is importantly analogous to standard divine attributes (more so, I think, than being a member of a trinity), and that triunity itself seems to be a great-making property, and thus a divine attribute (only triunity, not being a member of a trinity, seems to make interpersonal love relations internal to a being). By contrast, (1b) accommodates all of these considerations in one stroke. So one could be forgiven for preferring (1b) to (1c).

Can the best of (1b) and (1c) be combined? We might propose that deity includes the following disjunctive property, without entailing either disjunct:

1. (1d) Being triune or being a member of a trinity is included in the divine nature.Footnote ¹¹

The triune God has this property by satisfying the first disjunct, and the three members of the Trinity each have the property by satisfying the second disjunct. But on this proposal two things as different as a trinity of persons and a single person can both be divine, and yet two persons intrinsically just alike – perfectly similar except that one is a member of a trinity and one is not (e.g. a unitarian God) – will not both be divine.Footnote ¹² This makes deity seem like a very artificial and even gerrymandered property, but surely it isn't. Intuitively, deity is closer to the ‘joint-carvy’ end of the spectrum, maybe even an ‘Aristotelian’ natural kind.Footnote ¹³ Another worry is that (1d) may not do the work we want it to, at least not as well as (1b) does. For as I mentioned above, it is not (1d), but rather its first disjunct, that is importantly analogous to standard divine attributes, and is arguably a great-making property. And since it is more truth-functionally complex than (1b), a disjunctive property does not seem to preserve the same unified elegance in the concept of God that (1b) does. Finally, although (1d) can help secure a kind of explanatory connection between divinity and triunity, we are still left with an explanatory question similar to the one with which we began. An individual's satisfying (1d) entails that some other individual satisfies the other disjunct (there cannot be a trinity without members of a trinity, and vice versa), but there are a number of explanatory relations that might hold between the satisfaction of one disjunct and the satisfaction of the other. Is the first disjunct satisfied because the second is? Is the second satisfied because the first is? Are both satisfied in virtue of some third fact? It is not obvious what we ought to say here. I conclude that (1b) is, on the whole, better than its alternatives.

Clearly, denying (1b) comes at a price, and therefore reading (3a) ‘the Son is God’ as (3b) ‘the Son is divine’ does not by itself deliver us from trouble. Instead it simply exchanges one problematic triad for another, the second featuring (1b). Perhaps the consequences of rejecting (1b) are not devastating, but they are costly enough that an alternative, if it were available, would be welcome. For that reason, it is worth taking a look at other options before we bite this bullet.

Denying (2a)

In spite of initial appearances, maybe there is a palatable way to deny (2a), the claim that the Son is not triune. How might one try to deny this? The first idea that comes to mind is as unattractive as it is obvious: multiply persons in the Godhead beyond just three. Thus, we could say that the Son (as well as each of the other divine persons) consists of a further trinity of persons, resulting in many more persons in the Godhead than has traditionally been supposed. But this seems an enormous and inelegant departure from the traditional way in which the Christian God has been conceived. (Moreover, if every person in the Godhead is triune, then we end up with an infinite regress of divine persons, which is surely undesirable.)

There are subtler strategies we might try. Perhaps the Son is somehow composed of the other two divine persons (and something similar would have to be true of the Father and Spirit). It is at best unclear whether this is possible. Though I am not prepared categorically to rule this out, if there is an orthodox and plausible way to develop it, then, at the very least, much metaphysical labour lies ahead down this road. Up to this point, models of the Trinity have been designed to account for how a single God could be three persons, not how each single person could be, in some sense, three persons.

Here's another attempt. Latin Trinitarianism – paradigmatically Leftow (Reference Leftow2004) – has been interpreted in different ways. Some (e.g. Hasker (Reference Hasker2013) ) read Leftow as affirming that each of the divine persons is identical to God. If God is triune, and each person is (identical to) God, then each person is triune, and so (2a) is false.Footnote ¹⁴ The trouble with this move – and this is not news to defenders of Latin Trinitarianism – is avoiding modalism. If each person is identical to God, then each person is identical to each other person, and so, for example, the Father is identical to the Son. I think many will sympathize with Hasker's protest that this conclusion ‘is quite straightforwardly heretical’ (ibid., 116). So it is not obvious that this version of Latin Trinitarianism affords a palatable way out.

Others (e.g. Rea (Reference Rea, Flint and Rea2009b) ) read Leftow as claiming that the persons are distinct events involving a common substance, the events being God's life segments or life strands, the substance being God. On this reading, there is simultaneously a sense in which (2a) is true and a sense in which it can be denied. Strictly speaking, (2a) is true, for the Son is only one of God's three life strands, not the whole Trinity of them. But in a looser sense perhaps (2a) is false. Perhaps the Son is triune in a derivative way, simply by being a life segment of a God who is triune. (Presumably this is how it goes with the other divine attributes: the Son is omniscient, for example, simply in the sense that the Son is a life segment of a God who is omniscient.) And if the Son can possess triunity as well as all the other divine attributes in this derivative way, then alongside (1a) and a strict reading of (2a), we can also affirm something at least as strong as (3b)—that the Son is divine, possessing the full divine nature, albeit in a somewhat derivative way.

This is an attractive suggestion in a lot of ways, but there is still cause for concern. If the Son has triunity (derivatively) because triunity is a property of the God who lives the Son's life, then it seems the Son would also have all of the other properties of God at least derivatively, including the properties of living the Spirit's life and the Father's life, as well as the properties of performing the particular actions and operations unique to those lives. And the same goes for the other persons. This will result in unfortunate claims, such as that the Father (derivatively) has the properties of becoming incarnate and suffering. We will have to say that the Father has those properties in just as robust a sense as the sense in which he is divine. I do not think it is wholly clear whether this is sufficient for heretical theses like patripassianism, but we are close enough to such errors to be a bit uncomfortable. Moreover, if the sense in which the Father suffers (etc.) is weak enough to avoid heretical consequences, one might worry that the correspondingly weak sense in which the persons are divine is not robust enough to capture the spirit of the tradition. So in the end this proposal may not prove a success.

Denying (1a)

There appears to be no easy way out of (2a). But we still have (1a) to consider. Is there a tolerable way for an orthodox Christian to deny that God is triune? Maybe. The way to go about this is to identify God with each of the divine persons, and argue that, since no divine person is triune, neither is God. We've seen that one reading of Leftow (Reference Leftow2004) entails that God is identical to each of the divine persons, but we also saw that this interpretation of his view threatens to collapse into modalism, because it appears to entail that, e.g., the Father is identical to the Son. What the denier of (1a) needs is a way to block this consequence, and the most promising way to do that is to adopt a doctrine of relative identity.

On relative identity approaches to the Trinity, the sortal-relative identity claim ‘the Father is the same being as the Son’ is not reducible to ‘the Father is a being, the Son is a being, and the Father = the Son’. Moreover, the predicate __is the same being as__ does not ‘dominate’ the predicate __is the same person as__, so it is possible for the Father to be the same being as the Son but not the same person as the Son. According to relative identity models of the Trinity, each divine person is the very same being (or God) as each other divine person, but the persons remain distinct persons.Footnote ¹⁵

How do we get from here to a denial of (1a)? Take ‘the Trinity’ as referentially plural, picking out the Father, Son, and Spirit. God is not identical to this plurality, but only to each individual person. Yet no individual person is triune, so, strictly speaking, God is not triune.Footnote ¹⁶

This proposal has a lot to be said for it, but there are also reasons to worry. First, relative identity is not exactly popular; many philosophers reject it as ‘unintuitive, unintelligible, and uninformative’, among other things.Footnote ¹⁷ Of course, these concerns will not bother the philosophical theologian who already endorses a relative identity model of the Trinity on grounds independent of the Problem of Triunity, but even so, it would be better to have a solution to the Problem of Triunity that is not wedded to such a controversial notion.

Here is a second worry. Above I argued that there are reasons to think triunity is a property included in the divine nature. If that is true, then, a fortiori, it is one of God's properties, contrary to the present proposal. Can the relative identity theorist block my arguments that triunity is a divine attribute? She can block the explanatory argument by adopting (1c) in place of (1b). I claimed that the one and only God is a counterexample to (1c) because the one and only God is divine and yet not a member of a trinity. But on relative identity models that turns out to be false, since God is a member of the Trinity (Which member? All of them).Footnote ¹⁸ However, I also pointed out that (1c) does not accommodate the other reasons I gave in favour of (1b). Consider the appeal to perfect being theology. I sketched an argument that it is possible for something to be triune, and that a triune God would be greater than a unitarian God. So, according to perfect being theology, since God is maximally great, God is triune, contrary to the proposal under consideration. Let's consider one final objection to this argument. The relative identity theorist might point out that, if her model of the Trinity is correct, then presumably it is not just contingently true but necessarily true. So if God is not triune, then it is necessary that God is not triune. Therefore, even if it is possible for something to be triune (even if there could have been, e.g., a soul with three sets of cognitive faculties sufficient for personhood, or a multilocated rational substance that fulfils the person role three times over, etc.), it is not the case that God could have been triune. And if it is also true that nothing other than God could be omnipotent, omniscient, etc., then God might still be maximally great even if God would be greater if, per impossibile, God were triune.

This objection is related to a general problem for the method of perfect being theology that has been pressed by Jeff Speaks (Reference Speaks2014). The problem is roughly that, when we are reasoning about God's essential properties, ‘possibly God is F’ immediately entails ‘God is F’, leaving no space for the greatness of being F as opposed to not-F to do any work. Leftow (Reference Leftow2015) replies to Speaks by suggesting that the method of perfect being theology be understood in the following way. When we evaluate hypotheses about the divine nature, ‘we hold everything in the modal realm but theology constant, and the live options are just those that ‘fit’ into the modal realm as it actually is’, i.e. fit with ‘all non-theistic necessary truth’ (ibid., 428–429). So, if both ‘God is F’ and ‘God is not F’ fit with all non-theistic or non-theological necessary truth, then whether God is F or not F is decided by which of them it would be greater for God to be.Footnote ¹⁹

Now apply this to the question of triunity. My argument that God is possibly triune is effectively an argument that ‘God is triune’ fits with all non-theological necessary truth. ‘God is not triune’ also seems to fit with all non-theological necessary truth. So according to perfect being theology, whether or not God is triune is to be decided by which alternative would render God greater. I have argued that it would be greater for God to be triune than not triune. So the perfect being argument concludes that God is triune, contrary to the view under consideration.