Depicting Deity: A Metatheological Approach

Johnathan L. Kvanvig describes this book as an exercise in metatheology: an attempt to provide a framework for evaluating competing views about what is fundamental in theology. At the core of Kvanvig’s framework is the idea that ‘starting points’ for theologies ‘generate’ aspects of theologies, to which more must be added in order to arrive at adequate complete theologies.Kvanvig focuses on three starting points for theology, expressed as claims about divine essence:(CT) Fundamentally, a god is an asymmetric source of all else.(WWT) Fundamentally, a god is a being that is maximally worthy of supreme worship.(PBT) Fundamentally, a god is a supremely perfect being.Kvanvig takes these central claims to be, respectively, descriptive, evaluative, and normative.In order to assess the adequacy of these starting points, Kvanvig proposes two criteria: first, these starting points should ‘generate’ monotheism, personality, and (maybe) lack of embodiment; and, second, these starting points should ‘generate’ their competitors.The bulk of the book is devoted to an argument for the claim that, assessed against just these criteria, (PBT) comes in last, with (CT) either tied with or perhaps narrowly ahead of (WWT).Kvanvig does discuss a fourth option, which he calls ‘metatheological anti-fundamentalism’ (MAF). There are many forms that this option might take: perhaps a ‘combination’ of more than one of (CT), (WWT), and (PBT); perhaps a ‘subset’ containing more than one of (CT), (WWT), and (PBT); or perhaps even the view that no features of God are more basic or more fundamental than others. I think that, ultimately, Kvanvig evinces sympathy for something like the following view: (KV) Fundamentally, a god is an asymmetric source of all else that is maximally worthy of supreme worship.With this brief synopsis in hand, we turn to some of the worries that one might have about the views that Kvanvig develops.First, it is not clear that ‘metatheology’ is a good label for Kvanvig’s project. In particular, it seems odd in light of the analogy that Kvanvig tries to draw with metaethics. If we took that analogy seriously, we might suppose that the key question for metaethics concerns the choice between the following starting points:(C) Fundamentally, we should maximize good.(D) Fundamentally, we should act rightly.(V) Fundamentally, we should act virtuously.But, at least on many accounts, this choice is one of the central concerns of (normative) ethics rather than metaethics. And, at least on this analogy, it seems that what Kvanvig offers is squarely located in theology, rather than in anything that deserves to be called ‘metatheology.’Second, the idea that particular theologies have (potentially different) ‘starting points’ is one that admits of examination. Since Kvanvig draws explicit parallels between theological theorizing and scientific theorizing, it is worth recalling two prominent distinctions that philosophers of science draw: (i) the distinction between theorizing as process and theory as product, and (ii) the (possibly related) distinction between context of discovery and context of justification. It is obvious that, when we think about process and context of discovery, we may be thinking about starting points, but it is also obvious that those starting points need have no significance at all when it comes to product and context of justification. Equally, it seems clear that, when we think about product and context of justification, it is hard to point to anything that deserves the label ‘starting point.’ In principle perhaps, we might take the axioms in a best axiomatisation of a theory to be starting points; in principle, compactness of best axiomatisation is one dimension of the simplicity of a theory. However, given that Kvanvig acknowledges the holistic nature of the assessment of product and context of justification, it seems that there is no foothold for his kinds of starting points on the side of product and context of justification. The points that I have just made have some significance for some of the details of Kvanvig’s discussion. For example, he rules out:(PAT) God is pure act.as a candidate starting point for theology on grounds that invoke process and context of discovery. But those are insufficient grounds to refuse Thomism a seat at the table once we turn to product and context of justification. Kvanvig says that an approach that started with (PAT) would be as inadequate as an account that started with the claim that gold is the element with atomic number 79 (41). But, in fact, the claim that gold is the element with atomic number 79 is exactly the central claim to make in your theory about (the essential nature of) gold. Kvanvig also says that (PAT) is a ‘philosophical headscratcher’ (40) as a starting point, even if it makes sense in the theory in which it is embedded. But—all other considerations aside—this complaint has no interest in context of discovery and no force once we turn to context of justification.Third, the concepts of ‘generation’ and ‘derivation’ with which Kvanvig works also admit of examination, particularly in light of the distinctions mentioned in the preceding paragraph. Kvanvig takes pains to explain that, for him, ‘derivation’ can be either monotonic or non-monotonic. That is, when he considers what can be ‘derived’ or ‘generated’ from the claims that are the focus of his attention, he is not just concerned with the logical consequences of those claims, but also with what might be called ‘ampliative’ consequences of those claims. Set aside the thought that ‘ampliative consequence’ is merely enthymematic logical consequence. While it then seems right to think that, on the side of process and context of discovery, there is ampliative inference, it seems wrong to think that, on the side of product and justification, there is ampliative consequence. When we axiomatise, the only relation that we employ is logical consequence. Kvanvig notes a ‘holistic/atomistic’ distinction between ‘the conditions under which a [theory] is supported, confirmed or implicated by a given total system of information’ and ‘[the conditions under which] one has a reason for a given claim, a basis for inferring it, or some degree of confirmation for it’ (ix). But, while we have support, confirmation, implication, and reason in both contexts of discovery and contexts of justification, we have only bases for inferring in contexts of discovery. The distinction that Kvanvig draws crosscuts the distinctions needed.Alongside these rather high-level worries about Kvanvig’s project, there are also lower-level worries about some of the details of his arguments.Consider, for example, Kvanvig’s argument for the claim that monotheism clearly follows from (CT). As Kvanvig notes, at most one being can stand in the relation of being the asymmetrical source of all else to everything else. Moreover, according to Kvanvig, if we try giving a standard collective reading of the claim that the gods are the source of all else, we will locate the agency of the collective intentional action of the gods in the individual gods, so that the collectivity is agential only by courtesy, as in a musical performance by an orchestra. But if we are allowed to move to nonstandard collective readings—as, perhaps, in the Christian doctrine of the Trinity—then it becomes controversial whether the resulting view is really polytheistic. And if there is the possibility of conflict between the gods, then we face the Problem of the Pantheon, that is, the problem of reducing the exaltedness of deity to lamentable displays of the foibles and follies of human society. All of this argument is rather swift. To start with, we might wonder why we should prefer (CT) to: (CT*) Fundamentally, a god is a source of all else that has a source.Adopting this principle would block the immediate logical inference to monotheism. And, this early in the discussion, it is not clear why, for example, we should assume that it is not the case that, for example, numbers are independent necessary existents. More strongly, if we have any reason at all to think that there are numbers, then we have reason to suppose that they are independent necessary existents. For, on the assumption that there are numbers that God makes, we have to suppose—in my view, arguably incoherently—that, in the order of explanation, and perhaps even in the order of causation, prior to making the numbers, God was neither one nor many. Once we have come this far, we can go on to note that the further considerations that Kvanvig advances are not very impressive. There are many who think that, for example, institutions are not merely agential by courtesy: institutions have intentions that are not constituted from the intentions of their members, and so on. There are many who think that at least some widely accepted versions of the doctrine of the Trinity are polytheistic. There are also many who think that the so-called ‘Problem of the Pantheon’ is overstated: it is not too hard to suppose that a collection of very powerful, very wise, very good beings whose wills sometimes came into conflict would not exhibit the foibles and fallibilities of the members of, for example, the Greek pantheon.Here is another small example. Against Mark Johnston’s claim that, on the assumption that a separate creation has positive value, the value of the Highest One is less than the value of the Highest One plus a separate creation, Kvanvig objects that endorsement of this claim requires an inventive and non-Cantorian approach to value theory and the mathematics of infinite numbers (since, on a standard approach, the value of the Highest One is infinite, the value of the separate creation is finite, and the sum of infinite and finite is infinite). Setting aside pedantic worries (e.g., that, plausibly, on a Cantorian approach, the value of God will exceed all cardinal values), the important point to note is that the now not-so-recent literature on Pascal’s wager has already spawned a wide range of approaches that are able to make sense of the claim that Johnston makes. To give just one example, those who suppose that God and creatures cannot possibly lie on the same scale might think that value is vectorial: each value has the form ⟨d,t⟩, where d=divine value, c=creaturely value, d is either 1 (for the divine or a fusion that includes the divine), or 0 (for anything that does not have the divine as a part), and c is some real value, positive or negative. Values are lexically ordered: anything which has a 1 in the first slot of its vector trumps anything that has a 0 in the first slot of its vector, and, where things have the same values in the first slot of their vectors, higher values in the second slot trump lower values in the second slot. Finally, mereological fusions of separate things all of which have values themselves have values that sum over the values of their parts.