Divine Attributes and Their Tensions

The last two articles asked whether the attributes fit the world. This one asks whether they fit each other.

The question is prior to the other one and it is easy to skip. If the description is internally inconsistent, then nothing satisfies it, and no argument for or against the existence of the thing described is worth running. That is a stronger conclusion than any argument from evidence could reach, and it is available without leaving the armchair.

Where the list comes from

Anselm's formula from the ontological argument article is doing double duty. There it was a premise in an argument for existence. Here it is a research programme: settle what God is like by asking what it is better to be.

Thomas V. Morris gave the influential modern statement in Anselmian Explorations, University of Notre Dame Press (1987). The procedure is to take a candidate property, ask whether having it is greater than lacking it, and include it if so. Knowledge is better than ignorance, so omniscience. Power is better than impotence, so omnipotence. And so on down the list.

Two things about the method matter for everything that follows.

The first is that it runs on intuitions about what is greater, and those intuitions are contested. Is it greater to be capable of suffering with those you love, or to be beyond being affected by anything? The tradition answered firmly in one direction, and a good deal of recent work answers in the other, and there is no agreed way to adjudicate. William Wainwright and Peter van Inwagen have both pressed the point that without such a way the method cannot settle disputes, which is what it was for. Wainwright has also asked whether the being it constructs is the one anybody worships.

The second is that the method assembles a list without checking that the list can be jointly satisfied. Each property is selected on its own merits. Whether the selected set is coherent is a separate question, which the method does not ask, and this article is about what happens when someone asks it.

The form is worth noticing because it shares the structural advantage of the problem of evil. Everything the critic needs is supplied by the theist. No claim about the world is required, so no empirical dispute can arise, and success would establish not that God happens not to exist but that nothing could satisfy the description.

Two responses are always available, and they cost different amounts. One is to show the incompatibility is only apparent, by finding a reading of the attributes on which they cohere. The other is to revise or drop one of the attributes. The second is not automatically a defeat, since the tradition's formulations were not handed down complete, but it is a concession, and how much of one depends on which property gets revised.

Omnipotence, and what a puzzle is worth

Both answers look bad. If God cannot make the stone, there is something God cannot make. If God can make it, there is something God cannot lift.

The standard reply goes back to Aquinas and is about what counts as a task. Omnipotence is power over what is possible. A description of a logically impossible state of affairs does not describe a difficult task that God fails at; it describes nothing at all, and failing to bring about nothing is not a limitation. Making a square circle is not something God cannot do, because it is not a something.

George Mavrodes applied this directly in "Some Puzzles Concerning Omnipotence", The Philosophical Review 72, no. 2 (1963), pp. 221-223. If the being in question is omnipotent, then the phrase "a stone too heavy for that being to lift" is self-contradictory, in the same way as "a square circle". So the request to make one is not a request to do anything. C. Wade Savage's "The Paradox of the Stone", The Philosophical Review 76, no. 1 (1967), pp. 74-79, is the standard fuller treatment arguing that the paradox does not establish an inconsistency.

A more careful version distinguishes a being that is omnipotent necessarily from one that happens to be omnipotent and could cease to be. For the second, both making the stone and then being unable to lift it are straightforwardly possible, at different times, and the paradox dissolves without any appeal to what counts as a task. For the first, being non-omnipotent is impossible, so the stone is impossible, and Aquinas's answer applies.

Descartes dissented, apparently holding that God's power extends over the necessary truths themselves, so that the logically impossible is not a limit either. Almost no one follows him, and the reason is that the position cannot be stated: if contradictions can be true, the claim that God can make them true does not rule anything out, and an assertion that rules nothing out says nothing.

The stone is worth having as an example of the difference between a puzzle and a problem. It is answered, the answer is not ad hoc, and it survives in popular argument long after it stopped being live. A harder case in the same area is whether God can sin. If God is essentially good then God cannot, and there is then an act available to any ordinary person that God cannot perform. The usual reply is that being able to act wrongly is not a power but a liability, and that a being incapable of it is not thereby less able. That is a reasonable answer and it is not obviously a complete one, since the ability to do otherwise is precisely what the next section but one turns out to need.

Immutability

The motivation is easy to state and hard to escape. If God is perfect, any change is a change away from perfection. The argument was regarded as decisive for most of the tradition's history.

The cost arrives quickly. A God who cannot change cannot respond, cannot come to know that something has happened, and cannot be moved by anything, which sits badly with almost every religious text the doctrine was formulated to interpret. The tradition's answer involves a further set of doctrines about God's relation to time that the advanced tier examines in detail.

The sharpest form of the tension is Norman Kretzmann's, in "Omniscience and Immutability", The Journal of Philosophy 63, no. 14 (1966), pp. 409-421, and it is the model of this kind of argument.

An omniscient being knows everything that is true. Some things are true at one time and false at another: it is true today that a particular event is happening, and tomorrow it will be true instead that it has happened. So an omniscient being knows one thing today and a different thing tomorrow, and a being whose knowledge differs across times has changed. Omniscience therefore entails mutability, and immutability is inconsistent with it.

Kretzmann also pressed a second version that has proved harder. Some knowledge appears to be available only from a particular position: what a person in a hospital bed knows in knowing that they are the one in pain does not seem to be available to anyone else, however complete their information. If omniscience requires knowing everything anyone knows, an omniscient being would have to be each of them.

The responses divide along the lines the previous section set out. One approach reinterprets: God knows all truths in a way that does not involve occupying any moment, so there is no succession in the knowing and no change. That preserves both attributes at the price of a substantial account of God's relation to time, and it faces the question of whether a being outside time can know what is happening now. The other approach revises: give up strict immutability, allow that God changes in relation to a changing world while remaining constant in character. That answers Kretzmann directly and gives up something the tradition treated as settled.

Omnipresence

Augustine's formulation is that God is wholly present wherever God is present, and is not spread out by size in the way a body is, with one part here and another there. Anselm put the same point as a pair of claims that only sound contradictory: God is in no place, in the sense of being bounded by one, and is wholly in every place.

Aquinas gave the account that organised the medieval discussion, treating God as present by power, since everything is subject to God's action; by knowledge, since everything is open to God; and by essence, as the cause of each thing's existing at all. The presence is understood analogically rather than as literal location.

The interesting difficulty here is not an inconsistency but a suspicion. If omnipresence amounts to knowing what happens everywhere and being able to act everywhere, then it is not a further attribute at all. It is omniscience and omnipotence restated in spatial vocabulary, and the list is shorter than advertised. Aquinas's third channel, presence by essence, is what is supposed to prevent that collapse, and whether it succeeds depends on the account of causation behind it.

Recent work has taken the opposite route and argued that God is literally located, which raises the questions the classical view was avoiding: whether an immaterial thing can have a location, whether God and a chair can occupy the same region, and whether being everywhere requires having parts. Both directions are live, and this is one place where the standard list turns out to conceal a genuine open question rather than a settled doctrine.

Divine freedom

The tension is sharp and does not depend on anything controversial. Suppose in some situation one action is better than all the others. A being who is essentially perfectly good cannot do worse than best. So the being does that action, and could not have done otherwise, and on the most familiar account of freedom that is exactly what unfreedom is.

The problem generalises to creation. If God had reason to create, the reason applied necessarily, and a God who could not have refrained is not obviously the free creator of the tradition. If God had no reason, the choice was arbitrary, which is not much better.

Klaas Kraay's survey of the field, published in 2025, sets out four models of what divine freedom would require: being able to do otherwise, being free of external constraint, responding properly to reasons, and being the ultimate source of one's own actions. Only the first generates the problem in its acute form, and much of the response consists in arguing that the first is not what freedom requires. A being who acts from its own nature, unconstrained by anything outside itself, and perfectly responsive to reasons, has as much freedom as it makes sense to want, and the demand for available alternatives is a demand that the being sometimes act worse than it could.

Robert Adams argued in a different direction, that God may permissibly choose a good enough world rather than the best one, which restores alternatives at the cost of denying that a perfect being must maximise. The question of whether there is a best world at all, and what follows if there is not, belongs to the advanced tier, which takes it up on its own terms.

Next

These arguments were conducted without reference to anything that happens. The next article returns to evidence, and to the question of whether a report that something happened could ever make it reasonable to believe that a law of nature was suspended.