Omniscience and
Eternity: A Reply to Craig
Jonathan L. Kvanvig
William Craig presents an argument against those who maintain that there is no contradiction between being eternal and being omniscient. One of his targets is my own account of how an eternal God can be omniscient. Craig claims that my treatment of temporal indexicals such as ‘now’ is inadequate, and that my theory gives no general account of tense. Craig’s argument misunderstands the theory of indexicals I give, and I show how to extend the theory to give a general account of tense.
William Craig presents a limited argument against those who maintain that there is no contradiction between being eternal and being omniscient. The argument concerns what he terms “tensed facts,” facts concerning what McTaggart called the A-series regarding time. McTaggart distinguished the A-series regarding time, which has to do with temporal becoming and the classification of events in terms of past, present, and future, from the B-series, which concerns only the relations of simultaneity, before, and after among events. A B-theorist regarding time and tense maintains that all facts regarding the A-series can be reduced to facts regarding the B-series; an A-theorist maintains that the reduction goes in the opposite direction.
Craig’s argument is not meant to attack B-theorists, but is rather aimed at the remainder who still think it is possible for an eternal God to know all truths. Craig’s worry is about claims that report what is happening now, worrying that a being has to be in time in order to know what is happening now.
One of his targets is my own account of how an eternal God can be omniscient. There are two parts to his argument. First, Craig objects to my treatment of claims involving temporal indexicals, such as ‘now’. Second, he objects that my account is incomplete, since there is much more to the topic of tense than is covered by a theory of temporal indexicals. I will focus most of my attention on the first objection, and briefly show at the end of this paper how to extend my account to handle tense more generally.
Regarding my view of temporal indexicals, Craig’s objection concerns the two propositions
(1) It is now 1 June 1984
and
(1A) The essence of the moment picked out by the use of the demonstrative ‘now’ in (1) is mutually exemplified with the property of being 1 June 1984.
He says,
It is odd that Kvanvig refers to his (1) as a proposition, for on his own view propositions lack indexicality. Rather (1) is a sentence, and the question is whether the proposition expressed by (1) is infected by temporality, that is to say, whether that proposition is such that it cannot be expressed “without implying temporal indexicals.” Kvanvig’s claim is that the propositional content of (1) can be expressed by sentences not implying temporal indexicals. His reason for this claim is that (1A) is not so infected . . . and, moreover, contains all the same temporal elements as (1). But (1A) is true only if the “is” in (1A) is tenseless. Otherwise (1A) is false, having been true only on 1 June 1984. If (1A) is tensed, the time of its truth is just the same as that of (1). Thus Kvanvig errs when he says that if (1A) is not identical to (1) it is not because of some temporal element...[i]
Confusion abounds here, and it is partly my fault, for in formulating (1A) above, I referred to the word ‘now’ in (1), and that confuses the distinction between (1), a proposition, and the sentence
(S1) ‘It is now 1 June 1984’.
To remove this misleading element, we can reformulate (1A) as
(1A’) The essence of the moment picked out by the use of the demonstrative ‘now’ in (S1) is mutually exemplified with the property of being 1 June 1984.
This change does not remove all the confusions, however, and my goal here is to remove the confusions and show that Craig’s argument is not successful against the kind of theory I hold. I will outline the theory in question, and then return to Craig’s argument above.
My account of the compatibility of eternity and omniscience has nothing to do with tensed facts or with anything McTaggartian, except that I refuse to endorse a B-theory about time. In fact, one of the confusions that I think we need to eliminate is to quit talking about things like “tensed facts”. Tense is a property of sentences, not facts. Furthermore, on the theory I hold, propositions are the objects of knowledge, so if (1) above is something that can be known, it has to be a proposition. Craig thinks I must take it to be a sentence, but neither it nor (1A) can be taken to be sentences and still be objects of knowledge. Hence, when Craig inquires about the tense of (1) and (1A), he makes a category mistake. (1) and (1A) are not sentences, they are propositions, so they can’t be tensed.
Craig has an argument for thinking that I must be thinking of (1) and (1A) as sentences. He says that I hold that propositions “lack indexicality,” whereas (1) is formulated using the word ‘now’. Of course I hold that propositions lack indexicality. Anyone who distinguishes between sentences and propositions will hold the same, for indexicals are linguistic items, and as such, constituents of sentences, not propositions.
This response does little to clarify the theory I hold or how the relationship between (1) and (1A’) helps to show how it is possible for an eternal God to be omniscient in the face of the temporality of the world. So let me explain.
I assume, as already noted, that propositions are the objects of knowledge, and there are two importantly different theories of propositions. According to one theory, propositions are simply true or false; there is no such thing as a proposition that is true at one time and false at another. The other theory allows propositions to vary in truth value over time. I adopt the first kind of theory, holding that a proposition, if true at any time, is true at all times. Craig’s argument is not directed at the theory of propositions I endorse, so I will not enter into any defense of the view here.
I also distinguish between sentences and propositions, as also already noted, the former being linguistic items spoken or written which express the latter (when things go well semantically). So there is a difference between the sentence ‘We are now in the year 2000 A.D.’ and the proposition We are now in the year 2000 A.D., and difference which I will indicate by the practices just used of single quotes and bolded italics.
We can ask further questions about the nature of propositions, and the connection between sentences and propositions. Propositions are structurally complex, made up of metaphysical items which are the semantic values of the sentences used to express those propositions. When things go well for language, a sentence plus its meaning plus the context in question yields a single proposition as the semantic value of that sentence; when things do not go well, no such mapping exists. Things do not go well when the sentence is ambiguous, ill-formed, etc. Craig’s argument does not involve some slippage between sentence and proposition expressed, so the aspect of the theory of expression concerned with what happens when things do not go well are not important here.
Consider what happens when things go well, however. We have a simple sentence
‘John runs’
which, on a particular occasion of use, expresses the proposition
John runs.
I must note that the token formulation here is not intended to reveal the semantic structure of the proposition in question. I only use the token formulation as I do to signal the intimate connection between the proposition expressed and the linguistic structure of the sentence which expresses it. As a result, nothing can be inferred about the structure of the proposition from the token formulation I give of it. Moreover, another occurrence of that same token formulation need not represent that same proposition. So if there are two people named ‘John’ and someone asserts the same sentence above, I would use the same formulation of the proposition expressed as used above. But no one should infer that two tokens of the same type of formulation implies that I’m referring to the same proposition.
So if my manner of specifying the proposition in question above does nothing to reveal its structure, more work needs to be done. The particular structure of the proposition in question obtains in virtue of the semantic values of the terms of the sentence on the particular occasion of use. So, consider the following proposition
The semantic value of ‘John’ in the sentence ‘John runs’ has the property of running, which is the semantic value of the predicate of that same sentence.
Further, suppose that the semantic value of ‘John’ is some property which is the essence of John. Then we have the further proposition
The property which is the essence of John is mutually exemplified with the property of running.
If the semantic theory just employed is correct, then some sort of identification can be made between these three propositions. Just what kind of identification, I do not know. At the very least, the three propositions are necessarily equivalent, but the more interesting question is whether they are analytically equivalent to each other, or whether they are identical to each other. I hold no theory about that question, though I think it is a very interesting one. It is an interesting and deep question about semantic theory itself, about whether a proposition describing the semantic operations on elements of a sentence can be identical to the proposition expressed by that very sentence. Frege struggled with such a question, finding himself in the awkward position of claiming that the concept horse is not a concept.
So my primary aim is not to solve problems such as the above about the relationship between propositions and correct applications of a correct semantical theory. The only viewpoint I maintain about the relationship between the three propositions above is that the tense of the sentence used to express such the original proposition John runs is irrelevant to the question of the relationship between the three propositions. Whether the third proposition above is an accurate characterization of the constituents of the first is not dependent on issues of tense at all. What is important is that the second and third propositions are semantic elucidations of the first, telling us what metaphysical reality must be like for it to be true. In order for the semantic theory to be successful, it must correctly elucidate the constituents of the proposition in such a way that the proposition is true at all times if true at all. If it does so, then any worries about the tense of the sentence that expresses the proposition in question ought to disappear.
Similar remarks apply when the example is changed to one involving indexicals. For example, suppose on a particular occasion, I utter
‘I am tired’,
which expresses, on that occasion of use, the proposition
I am tired.
The assumed semantical treatment of the sentence used on that occasion yields
The semantic value of ‘I’ in the sentence ‘I am tired” has the property of being tired, such property being the semantic value of the predicate of said sentence.
If we adopt a semantical theory according to which the semantic value of first-person indexicals is the essence of the speaker, the semantic theory yields
The essence of Jon Kvanvig is co-exemplified with the property of being tired.
Once again, no issue of tense arises here, even though the sentence used to express the propositions in question is a present tense sentence. What matters, to repeat, is the adequacy of the semantical theory that proposes the second and third propositions as semantic elucidations of the first. That theory must delineate the constituents of the proposition I am tired in such a way that the proposition is true at all times if true at all. If the theory is able to do so, then worries about indexicals for the theory of propositions I hold disappear.
One may complain that the propositions in question are not fully represented, because they fail to specify the time at which I am tired, and one may choose to expand the example to accommodate this desire for full representation. We may do so by considering what is expressed by the sentence
‘I am now tired’.
It is the proposition
I am now tired.
Using the same treatment outlined above, our semantical theory yields
The semantic value of ‘I’ in the sentence ‘I am tired” has the property of being tired now, such property being the semantic value of the predicate of said sentence.
If we adopt a semantical theory according to which the semantic value of first-person indexicals is the essence of the speaker, the semantic theory yields
The essence of Jon Kvanvig is co-exemplified with the property of being tired now.
We may inquire further into the constitution of the property in question, the property being tired now. Since our propositions are true at all times if true at all, we should adopt a semantical theory that treats this property as complex, composed out of the property of being tired and some appropriate semantical treatment of the term ‘now’ as used in the sentence above. By analogy with the treatment of the indexical ‘I’, we can adopt a theory which makes the semantic value of ‘now’ the essence of the moment of utterance of the sentence containing it. Such a treatment makes the property being tired now a complex entity composed of the property of being tired together with the essence of the time of utterance. Such a theory presumably would treat the entire proposition as involving the two-place relation being tired with the essence of Jon Kvanvig and the essence of the time of utterance as its two relata.
I stress again that issues of tense play no role in such a theory. At the risk of boring the reader, what is at issue is the semantical theory which posits certain semantic elucidations of propositions expressed by tensed sentences involving indexicals. Should there be no such adequate theory of semantic constitution for handling issues of tense and temporal indexicality, then the theory of propositions presupposed must be abandoned and some other approach would be needed to reconcile God’s eternity and omniscience. If no such objection is forthcoming, then the semantical theory in question gives some grounds for thinking that eternity and omniscience are not incompatible. For the theory in question purportedly delineates the constituents of the proposition I am now tired in such a way that this proposition is true at all times if true at all, and hence worries for the theory of propositions that I hold based on considerations of indexicality and temporal indexicality in language disappear.
This theory is not a complete theory of propositional attitudes, for it cannot explain the difference, for example, between de se and de re belief.[ii] In order to complete the account of propositional attitudes, a third relata must be included in addition to the person and the proposition in question, what we might refer to as a way of accessing, or a mode of presentation, of the proposition in question. According to this theory, the difference between de se and de re beliefs that have the same propositional content is found in the way of accessing that content. Similarly, the difference between a temporal knowing that it is now noon and an atemporal knowing of that same propositional content is found in the way of accessing that content. I will not pursue the details of the theory here any further, since Craig’s objection to my theory does not concern this third element of propositional attitudes, but focuses instead on the propositions involved. I turn then to that objection.
Craig’s central point is contained in the following passage: “But (1A) is true only if the “is” in (1A) is tenseless. Otherwise (1A) is false, having been true only on 1 June 1984. If (1A) is tensed, the time of its truth is just the same as that of (1).”[iii]
Again, I must insist that matters of tense have nothing to do with the issue. What is relevant is whether the propositions in question are true simpliciter, or whether they vary in truth value from one time to another. Craig believes that (1) does so vary, and hence that (1A) must also so vary in order to identified with (1). But why does Craig think that (1) changes truth value across time? As far as I can tell, the only reason would be because the sentence which expresses (1) expresses a truth at some times but not at others. But (1) is not a sentence; it is a proposition. So properties possessed by the sentence that expresses it on a particular occasion of use should not automatically be thought to be properties of (1) itself.
Craig might claim perplexity at this point: how could (1) be a proposition of the sort consistent with the view of propositions I hold, i.e., how could (1) be a proposition that is true simpliciter? The answer is this: that is the point of employing the semantic theory which generates (1A’) as the semantic elucidation of the structure of proposition (1). (1A’) tells us what are the metaphysical constituents of (1), thereby revealing how (1) can be true simpliciter. One gets confused about what my theory is proposing by failing to distinguish (1) from the sentence which expresses it, and failing to note that (1A’) is the semantic elucidation of what is involved in (1)’s being true at all times (and also by failing to recognize that other tokens of ‘It is now 1 June 1984’ need not specify the very proposition under discussion, for the proposition under discussion is that one with the constituents elucidated by (1A’)). Without the semantic elucidation, it would be mysterious indeed to claim that (1) is true at all times, since the terminology used to formulate it certainly looks like the kind of terminology that would be used to express a claim that is true at some times but not at others. But with the elucidation, any objection to this attempt to reconcile eternity and omniscience must claim that the semantic theory is problematic. It is interesting to note that Craig does not do that.
I wish to urge some caution about how successful a reconciliation I take this to be. First, I don’t believe there is a lot at stake here, because I, along with Craig, do not think the doctrine of the eternity of God is well-motivated. Furthermore, I am not convinced that moments of time have essences. It could be that moments of time are purely relational, having no identity in themselves other than the place they hold in the temporal sequence, or, more radically, it could be that the concept of an essence is simply incoherent. I know of no demonstration of either of these possibilities, however, and in the absence of such an argument, the above approach suggests that eternity and omniscience are compatible.
One other item remains, for the approach I cite applies only to sentences containing indexicals and not to tensed sentences generally. But it is not hard to extend the treatment to apply more generally. Some claims about the past and future involve demonstratives such as ‘yesterday’, ‘tomorrow’, ‘one week hence’, etc. I would propose to treat these similarly to the indexical ‘now’: what is expressed is the essence of the moment referred to. Besides these kinds of sentences, there are past and future tense sentences, which can be treated as quantifications over times that have the properties of having existed in the past or future, respectively. That leaves the present tense sentences as the difficult case, but I see no impediment to holding that a present tense sentence such as ‘I am tired’ expresses the same proposition as the sentence ‘I am now tired’. I am sure there are complexities to be addressed, such as problems about tenses that signal continuous action in the past, present, or future, but I see no reason to think that any such problems would be insurmountable. If that is correct, then the treatment of temporal indexicals outlined above can be extended to show how to reconcile temporality in all its aspects with the omniscience of an eternal God.
So, in short, Craig’s objection is an objection to a theory quite different from the one I defend. Perhaps there are reasons to abandon the approach I suggest, and if so, I side with Craig by retaining the doctrine of omniscience and abandoning the doctrine of eternity. Still, abandoning the latter doctrine would be premature at this point, since Craig’s worries about co-tenability do not extend to the account of such outlined here.
Texas A&M University
Notes