Analytic Theology, et cetera.

Home » Analytic Theology » Jonathan Edwards and Necessity

Jonathan Edwards and Necessity

I. Introduction

A friend wondered if I could say something more about the charge that Edwards commits a modal fallacy—in this case, it is alleged that from 1. necessarily, if α then β, 2. α, he concludes, 3. necessarily β—in the course of his argument for determinism (see this post for context). Specifically, he wondered if I might cite more from Edwards. In this post I’ll quote one of Edwards’ arguments for the necessity our actions have, and his reasoning should make clear that the charge leveled by some—namely, Richard Muller, some associated with the Uterecht school, and (some of) their students—is simply not viable. Some of this will be a repetition of my last post, but I view what follows as a more decisive response to Muller et al., than my previous post.

II. Conceptual Toolkit

Here’s some background. Where we let ‘Nec‘ represent ‘necessarily’; ‘α’ and ‘β’ represent propositions ; and ‘→’ represents ‘if α then β’, then,

  • Nec(α → β)

Statements of this form may be said to represent the necessity of the consequence. Here we’re saying that ‘α’ cannot obtain without ‘β’ also obtaining. ‘α’ and ‘β’ are connected by a necessary connection. Here’s an example from (more or less) natural language

  • Necessarily, if Smith is a bachelor, then Smith is unmarried.

Thus, in all situations where Smith is a bachelor, Smith is also unmarried in those situations. There’s never a situation where Smith is a bachelor but happily married (or unhappily, for that matter!).

Now, take a sentence with a different structure,

  • α → Nec(β)

Statements of this form may be said to represent the necessity of the consequent. The necessity isn’t on the joining of α with β, but on β itself. We’re saying here that if α happens to obtain, then β itself is necessary—that is, β obtains in every situation. Let’s explore this via the above case of Smith’s bachelorhood. We might say,

  • If Smith is a bachelor, then he must be unmarried.

This sentence is ambiguous. If you meant it to have the structure of our first sentence designated by the bullet point, you’d be correct. But suppose you really meant it to express the structure designated by our third bullet point, i.e., ‘α → Nec(β)’. Then you’d have a problem. Here’s how:

Suppose you agree with these two premises

    1. Necessarily, if Smith is a bachelor, then Smith is unmarried.


    2. Smith is a bachelor.

What you would be entitled to conclude is,

    3. ∴ Smith is unmarried.

but not,

    3*. ∴ Necessarily, Smith is unmarried.

What (1) says is true in every world. So let’s pick an arbitrary world, call it n. Now, suppose that in n, Smith is a bachelor at time t. We must conclude that in n Smith is unmarried at t. That’s (3). If instead of (3), (3*) followed, we would then be saying that in every world, Smith is unmarried. That is, from the mere fact that Smith is a bachelor at t in n, it follows that Smith is unmarried in every world, l, m, o, p, q, . . ., ad infinitum. But surely that doesn’t follow. Think of all the bachelors you knew in our world who are now married. Their being a synchronic bachelor didn’t entail their diachronic bachelorhood.

Okay, so let’s see the form of the valid syllogism above, the one that moves from (1) and (2) to (3). Where Bx = x is a bachelor, and Ux = x is Unmarried, and s = Smith, then

    1. Nec(Bs → Us)
    2. Bs
    3. ∴ Us

This is valid reasoning. No fallacy is committed here. The invalid form that Edwards is accused of employing, the one that illicitly transfers the ‘Nec‘ to the consequent, looks like this,

    1. Nec(Bs → Us)
    2. Bs
    3*. ∴ Nec(Us)

This isn’t good reasoning. Reasoning that takes this form commits the modal fallacy of transferring the necessity of the consequence to the necessity of the consequent. This is the form of reasoning Jonathan Edwards is accused, by Muller et al., of employing in Freedom of the Will (FOW).

Before moving on to what Edwards says in FOW, it will be prudent to put one more form of reasoning into our toolkit. There is a way to validly get premise (3*) above. Take a look at the Distribution Axiom in modal logic, which states,

  • Nec(α → β) → (Nec(α) → Nec(β))

Don’t let that scare you! The distribution axiom says that if it is necessary that if α then β, then if necessarily α then necessarily β. We could get (3*) above if we attached a ‘Nec’ operator to ‘Bs’ in premise (2) above. The valid form would look like this:

    1. Nec(Bs → Us)
    2*. Nec(Bs)
    3*. ∴ (Us)

This chain follows the distribution axiom, so we’re good.  Thus, while the inference from (1) and (2) to (3*) is invalid, the inference from (1) and (2*) to (3*) is valid. Note also that we could not get (3*) without (1), so we need the necessity of the consequence in such an argument. Of course, we would doubt the truth of (2*), but that doesn’t change the validity of the argument. Someone who makes an argument instanced by the general form (1), (2*), and (3*) instances, does not reason fallaciously. At best, he’s only guilty of making an unsound argument. This last point is important. We will make some adjustments to the operator used on the propositions in (2*) and (3*), but the basic idea will remain the same. That idea is that there are valid ways to transfer necessities to the consequent.

III. Jonathan Edwards and Muller & Co.’s Charge

Commenting on a statement in Jonathan Edwards’ FOW, Richard Muller states,“Edwards, it should be noted, appears to confuse [necessity of the consequence] with [necessity of the consequent]: he uses the former term but clearly means the latter” (“Jonathan Edwards and the Absence of Free Choice,” Jonathan Edwards Studies vol. 1, no. 1 (2011), 14).

So the claim here is that, though Edwards explicitly states that he’s talking about the necessity of the consequence, Muller claims Edwards does not mean what he explicitly says he means, and he clearly does not mean it. Rather than look at the two sentences from Edwards that Muller pulls from a larger section, and then makes his allegation that Edwards clearly means something different from what he explicitly says he means, I suggest we look at a larger, and more perspicuous quote from FOW.

In the following, Edwards explicitly mentions the necessity of consequence and claims all of our actions are necessary. Surely this would be a prime candidate to ferret out an instance of when he commits the relevant modal fallacy we’re concerned with. Thus Edwards,

I come now, in the second place, to show the consequence; how it follows from hence, that these events are necessary, with a Necessity of connexion or consequence.


1. I observed before, in explaining the nature of Necessity, that in things which are past, their past existence is now necessary: having already made sure of existence, it is too late for any possibility of alteration in that respect; it is now impossible that it should be otherwise than true, that the thing has existed.

2. If there be any such thing as a divine Foreknowledge of the volitions of free agents, that Foreknowledge . . . long ago had existence; and so, now its existence is necessary; it is now utterly impossible to be otherwise, than that this Foreknowledge should be or should have been.


3. It is also very manifest, that those things which are indissolubly connected with other things that are necessary, are themselves necessary. As that proposition whose truth is necessarily connected with another proposition, which is necessarily true, is itself necessarily true. . . (FOW, p.60)

Notice that Edwards clearly says that he’s concerned with “a Necessity of . . .consequence.” Notice further that Edwards says that “these events,” e.g., the putative free actions of humans, are necessary. This is precisely what bothered Muller in the quote he cited from Edwards, but here it’s even more clear. In the quote Muller used, Edwards merely “inquired” whether putatively contingent human actions could be divorced from “all Necessity.” Here, he asserts that they cannot be, and then endeavors to show that they cannot be. Surely this, rather than the two sentences Muller quotes, is a more relevant and interesting passage from which to discern whether or not Edwards commits the relevant modal fallacy in question.

III.a Edwards’ reasoning in the above passage

Ad premise (1): Here Edwards is appealing to the accidental necessity of the past. He says that once a thing is over and done with, it’s too late to now alter it. It is now impossible to make it the case that the thing that was true, or did exist, wasn’t true or didn’t exist. Edwards is basically saying here that there’s no crying over spilt milk (HT: Linda Zagzebski). If you spilled a glass of milk earlier today, there’s nothing you can do now to un-spill it.

Ad premise (2): Here Edwards claims that divine foreknowledge of the “volitions of free agents” is something that existed in the past and so is something we cannot now do something to change. Let’s make this more concrete. Edwards is saying that, since God is essentially omniscient, then yesterday God infallibly believed that you will spill a glass of milk earlier this morning. it’s now impossible to make it the case that God didn’t believe that proposition yesterday. Think of your own beliefs. Suppose that yesterday you believed that it was a cold day for mid-March. Is there something you can do today to make it the case that you didn’t in fact believe what you in fact did believe yesterday? It doesn’t seem so.

Ad premise (3): Notice here that Edwards is basically stating our distribution axiom from above! Edwards says, “It is also very manifest, that those things which are indissolubly connected with other things that are necessary, are themselves necessary.” It looks like Edwards is affirming that: Nec(α → β) → (Nec(α) → Nec(β)). His conclusion that our “voluntary free actions” are “necessarily true,” is because they are connected by necessary connection (necessity of consequence) to antecedents that are necessary.

So let’s give an initial construction of his argument. Edwards, with the rest of Reformed theology, affirms that God’s free knowledge of what men will do is based on his decree. So if God foreknows that p, then God has decreed that p. Where G = God knows that p and p= some proposition about what a free creature will do, then

  1. Nec(G → p)
  2. Nec(G)
  3. ∴ Nec(p)

There’s no modal fallacy here! You may disagree with Edwards. You still may claim he’s departing from the Reformed tradition. But you may not, any longer, claim that his argument confuses the necessity of the consequence with the necessity of the consequence. Or at any rate, you may not if you wish to remain intellectually honest—and we’ll suppose you do.

However, we’re actually not quite done yet. The above argument (III.a 1–3) is problematic in several ways. There’s some wrinkles to iron out and some salient points worth making. I’ll turn to that now.

III.b A wrinkle to iron out: types of necessity

As I’ve been using it, ‘Nec’ stands for something like logical or perhaps metaphysical necessity. In my previous post, I said if some sentence was scoped by ‘Nec’, that means it is true in all situations, where a ‘situation’ means something like ‘all ways the world could have been’. So if we let the ‘p’ in premise (3) in the above III.a argument stand for “Gillian will spill the milk at time t,” then this means that in every possible situation, Gillian spills milk at t. In this sense, we might say that Nec represents absolute necessity. A thing is absolutely necessary just in case it is true in all possible worlds. In Edwards’ time and before, something that was absolutely necessary was necessary in itself. It is the necessity possessed by a truth T such that the denial of T leads to a contradiction (e.g., denying that 1+1=2). It does not seem that denying “Gillian spills milk at t” leads to a contradiction—indeed, it would place severe limitations on God to say that he was unable to create a world where Gillian doesn’t spill the milk at t. This seems unorthodox. So, this type of necessity won’t do. Even though the above argument in III.a doesn’t commit the relevant modal fallacy, it would have been denied by Reformed theologians prior to Edwards; and, I might add, it should be denied by post-Edwards Reformed theologians too!

We are not back to square one, however. The Edwardsian argument listed in III (1)–(3) doesn’t make use of absolute necessity, but, as was stated previously, something more like accidental necessity. Edwards himself recognized several types of necessity, and early on in FOW he went to great pains to distinguish them and to say, precisely, how he understood the type of necessity that attached to free acts of the will. It is worth quoting Edwards at some length here on some of the ways something can be necessary:

(1.) They may have a full and perfect connection in and of themselves; because it may imply a contradiction, or gross absurdity, to suppose them not connected. Thus many things are necessary in their own nature. . . . So God’s infinity and other attributes are necessary. So it is necessary in its own nature, that two and two should be four; and it is necessary that all right lines drawn from the center of a circle to the circumference should be equal. It is necessary, fit and suitable, that men should do to others, as they would that they should do to them. So innumerable metaphysical and mathematical truths are necessary in themselves: the subject and predicate of the proposition which affirm them, are perfectly connected of themselves. [PM: absolute necessity]

(2.) The connection of the subject and predicate of a proposition, which affirms the existence of something, may be fixed and made certain, because the existence of that thing is already come to pass; and either now is, or has been; and so has, as it were, made sure of existence. And therefore, the proposition which affirms present and past existence of it, may by this means be made certain and necessarily and unalterably true; the past event has fixed and decided the matter, as to its existence; and has made it impossible but that existence should be truly predicated of it. Thus the existence of whatever is already come to pass, is now become necessary; it is become impossible it should be otherwise than true, that such a thing has been. [PM: accidental necessity]

(3.) The subject and predicate of a proposition which affirms something to be, may have a real and certain connection consequently; and so the existence of the thing may be consequently necessary; as it may be surely and firmly connected with something else, that is necessary . . . either fully and thoroughly connected with that which is absolutely necessary in its own nature, or with something which has already received and made sure of existence. This Necessity lies in, or may be explained by, the connection of two or more propositions one with another.——Things which are perfectly connected with other things that are necessary, are necessary themselves, by a Necessity of consequence.

And here it may be observed, that all things which are future, or which will hereafter begin to be, which can be said to be necessary, are necessary only in this last way. Their existence is not necessary in itself; for if so, they always would have existed. Nor is their existence become necessary by being already come to pass. Therefore, the only way that any thing that is to come to pass hereafter is or can be necessary, is by a connection with something that is necessary in its own nature, or something that already is, or has been; so that the one being supposed the other certainly follows.—And this also is the only way that all things past, excepting those which were from eternity, could be necessary before they come to pass; and therefore the only way in which any effect or event, or any thing whatsoever that ever has had or will have a beginning, has come into being necessarily, or will hereafter necessarily exist. And therefore this is the Necessity which especially belongs to controversies about the acts of the will. (FOW, pp. 16-17, emphasis supplied)

Here Edwards tells us, very early on, exactly what he means by “a Necessity of consequence.” It is things that are perfectly connected with things that are necessary in sense (1) or (2) above. Note further what he includes as examples of necessity in (1)—i.e., absolute necessity. He includes truths of mathematics, logic, (certain) moral principles, etc. These things, per Edwards, are unconditionally necessary, and hence, necessary in themselves. The type of necessity at issue in (2) is accidental necessity. This includes those things that have happened or are presently happening. Then, in (3), Edwards tells us that the only way in which a future free action is necessary is if it is “firmly connected with something else, that is necessary . . . either . . . absolutely necessary in its own nature, or with something which has already received and made sure of existence.” That is, free actions inherit their necessity from the antecedent’s necessity. And the antecedent may be necessary in one of two ways: (a) absolutely necessary, or (b) accidentally necessary.

An example of the first type might be something like this:

  1. Nec( if 2+2=4, then Gillian will spill milk at t.)
  2. Nec(2+2=4).
  3. ∴ Nec(Gillian will spill milk at t.)

It is unclear whether Edwards thought any of our actions inherited their necessity in this way. If there are any at all, they’re very few and far between. Edwards does seem to say that the vast majority, perhaps all, of our actions do not inherent their necessity from ‘Nec’, but from accidental necessity. Recall that in the quoted section in III.b. (3) above, Edwards stated that, “this also is the only way that all things past, excepting those which were from eternity, could be necessary before they come to pass.” So, those past things that are necessary and connected with a future free action are themselves merely accidentally necessary. Therefore, they will not be scoped with ‘Nec’, but with some other suitable operator.

Now, it is well-known that Edwards held to classical compatibilism. As he saw it, our free actions were determined by our preceding strongest desire(s) or motive(s) (a more rigorous account would spell out, among other things, what is meant by ‘strongest,’ but I will unfortunately forego such desiderata in this post). Now, per Edwards above, that Smith has psychological desire D at t is accidentally necessary for all t*’s ≥ t (surely Edwards wouldn’t say D is “from eternity”!). Something that is accidentally necessary is, per Edwards, now “unalterably (or unpreventably) true.” Let Unp stand for ‘unpreventably’. Then Edwards’ argument would have this form:

    4. Nec(if Smith’s strongest desire at t is for X, then Smith will X at t3).
    5. At t2, Unp(Smith’s strongest desire at t is for X).
    6. ∴ At t2, Unp(Smith will X at t3).

It would take more than a half-dozen premises to lay out the assumed premises of this argument, but for present purposes what matters is that, per Edwards, the above (4)–(6) is an example of an Edwardsian argument for the necessity our future free actions have. He runs similar arguments from foreknowledge. This isn’t to overdetermine our actions in terms of necessities, but to show in various ways that our actions are necessary, in the specific kind of necessity—i.e., not absolute necessity—that Edwards indexes to them. The upshot is that (4)–(6) does not commit the modal fallacy Muller & Co. accuse Edwards of committing. I hope to have shown that charge is no longer viable. Indeed, in God, Time, and Freedom (1989), William Hasker shows similar arguments like (1)–(6) directly above are found in Aquinas (pp. 8–11). Aquinas says things very similar to Edwards. Are we to suppose that Muller would also charge Aquinas with committing basic logical blunders, as he has Edwards?

III.c  Two final points

I would now like to make two brief points that bear on Edwards’ argument, and his metaphysics more broadly.

c.1 God’s past beliefs are not accidentally necessary because he’s timeless

To be fair to Edwards, he has many ways of getting his determinist conclusion. Showing that God’s past beliefs are not accidentally necessary would not, therefore, make one scot-free, and for at least two reasons: (a) Edwards has other arguments for determinism, and (b) the timeless realm seems just as fixed and unalterable as the past. To quote Zagzebski again: “If there’s no crying over spilt milk, there’s no crying over timelessly spilt milk!” Nonetheless, I don’t have the time to argue these points. For even granting it would not show that Edwards committed a modal fallacy, and that’s what I’m concerned with herein.

c.2 Edwards held to the necessity of creation; in fact, to the necessity of this creation

This is an interesting issue. First, what is the relevant sense of ‘necessity’ involved? One might think it is moral as opposed to absolute necessity. Here’s an argument against the idea that our world is absolutely necessary. Aside from Edwards’ suggestions that our world, or at least the free actions therein, are accidentally necessary, recall that for Edwards, and in accord with the tradition, a thing or proposition is absolutely necessary if its denial would lead to a contradiction. Now, Edwards did think that creation was a free act of God’s. On the view in question in this section, this free act would be compatibilist free. Here’s how the Stanford Encyclopedia of Philosophy describes Edwards’ view of God’s freedom here:

[God] is nevertheless free in the sense that he is aware of alternatives (the array of possible worlds), has the ability (i. e., the power and “skill”) to actualize any of them, is neither forced, constrained nor influenced by any other being, and does precisely what he wishes.

Note that God, on this view, is “aware of the alternatives (the array of possible worlds), [and] has the…power to actualize any of them.” For God to be aware of an alternative, and have the ability to bring it about, then he’d have to deny what is actual. But if what were actual were absolutely necessary, and on the supposition that God is unable conceive of or bring about contradictions, then, for Edwards, creation isn’t absolutely necessary. However, even if it were, the argument for the necessity of our actions would have the form of the distribution axiom, and thus, while such a view is false, it isn’t fallacious. And that, again, is what I am concerned with herein.

Thus, unless convinced otherwise, I conclude that Edwards is innocent of the oft-leveled charge by Richard Muller & Co., viz., that he confuses the necessity of the consequence with the necessity of the consequent. The moribund charge is now dead.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

March 2014


Follow Analytic Theology, et cetera. on
%d bloggers like this: