I. FOREKNOWLEDGE

Although many events are never known to occur until they actually happen, a person might believe that, in some sense, these events could have been foreknown. A determinist, some have said, is one who believes that all events can be foreknown.

This definition is adequate, of course, only insofar as an adequate analysis of knowledge-statements is at hand. There is, however, a great deal of disagreement about such an analysis. Are knowledge-statements descriptive or performatory? Do they entail belief-statements? How much and what kind of evidence must a person possess before he may be said to know something?

One may naturally feel that if it is foreknown by A that B will perform a certain action or make a certain decision, then B does not really act or decide freely. Although the relationship between foreknowledge and free will was discussed traditionally in a theological context, e.g., by St. Augustine and Hobbes, we need not be concerned about the characteristics A possesses in addition to his having this particular foreknowledge unless the nature of A's foreknowledge is distinctive.

People have believed that divine foreknowledge is different in kind from the human variety. But if we allow two kinds of foreknowledge, human and divine, what implications does this have for a definition of determinism? Assuming that a definition in terms of foreknowledge is feasible, we might try: All events can be foreknown by humans and are foreknown by God. (Since all God's potentialities are fully actualized, we do not want to say merely that He can know everything that will happen.) We shall discard this implication, however, because the conception of determinism in which we are interested does not presuppose theism; moreover, we should have to clarify further the nature of divine foreknowledge. In this context, it is interesting to note that St. Augustine's arguments in defense of the compatibility of free will and God's foreknowledge (Bk. m, sees. 2, 3, and 4) do not presuppose a unique conception of foreknowledge for God.

In order to avoid a commitment to theism, we might change the "and" to "or" in the definition and be able to read the second clause as: If there were a God, He would foreknow everything that happens. Determinism might then be a necessary truth for it would be a disjunction, and one disjunct might be necessarily true. "If there were a God, He would foreknow everything that happens" is necessary if fore-omniscience is included in the divine essence. To discover whether or not it is a necessary truth, one would have to get deeply embroiled in theology.

The truth of the proposition in question may be grounded in God's atemporal character. That is, since God is outside time, one cannot specify the time at which He knows something. But then one should not speak of "divine foreknowledge," since God does not know that something will take place at t, prior to t: from His vantage point in eternity, the whole temporal process is "non-temporally before God's eyes." If this conception can be rendered intelligible, it would appear that no incompatibility exists between divine foreknowledge and freedom since God knows what Smith will do in a way very similar to that in which Jones, who is watching Smith do it, knows. The problem of foreknowledge requires that we suppose the knowledge to exist prior to the event known; for it is only when this condition is satisfied that we become concerned about our freedom. And we have adopted a desideratum for a definition of determinism according to which determinism and freedom must be prima facie incompatible. Divine omniscience disturbs us because we view God as having already observed us doing things that we have not yet done. If, however, we are precluded from applying temporal predicates to God's epistemological states, then we cannot produce the formulation that disturbs us and must look elsewhere if we wish to satisfy this desideratum.

One might try to ground the proposition in the fact that God makes everything happen and, thus, has foreknowledge in a way similar to the way I know how the chair I am constructing will turn out. But one cannot be concerned about freedom unless one supposes an actual God who makes us do things. That is, no one worries about the freedom of his future actions because he accepts the proposition "If there were a God, He would make us perform the actions we do perform" unless he also accepts the antecedent of this conditional. Again, our desideratum is not satisfied.

Thus, if a person does not want to assume theism, but wants to know why foreknowledge seems to threaten freedom, he obviously cannot cite properties of a hypothetical God that are linked to divine foreknowledge. We may, though, cite properties of the world that, it may be argued, are related to His foreknowledge. For example, one may say that if there were a God, He would foreknow all that happens because the world is deterministic and He would have all the relevant information. Ignoring the fact that we are supposed to be defining determinism in terms of foreknowledge, it is clear that our concern about freedom arises from the assumption that the world is deterministic, not from the fact that there might be a God who knows the true deterministic theory. Hence, I see no point in referring to divine foreknowledge in the definition of determinism. I further contend that our concern about freedom does not arise from the fact that there might be someone, human or divine, who knows the true deterministic theory.

Libertarians emphasize the distinction between free and voluntary actions, insisting that an action may be voluntary and yet not free. If a voluntary action is, roughly, one I perform but am not constrained to perform, they wish to insist that the absence of constraint is, at most, a necessary condition of freedom. It would appear that a person who believes that foreknowledge and freedom are incompatible would be forced to accept this libertarian point. For it would be odd to conclude that B is constrained to do something merely because someone else knows that he will do it. Examples bring this point home. If I know that you will give money to a certain charity because I know you are generous, are sympathetic with this particular charity, and are not at the moment pressed for funds, it is very odd to conclude that you will be constrained to give the money as if at the point of a gun. A person who believes that foreknowledge and freedom are incompatible, therefore, would presumably have to argue, like the libertarian, that voluntariness is not a sufficient condition of freedom.

But, in fact, some of the persuasiveness of the doctrine that foreknowledge and freedom are incompatible derives from a conception of the relation between the foreknower and the act foreknown that is very much akin to constraint. Why should one have this conception?

Some who have proposed an analysis of knowledge-claims assert that "A knows that B will do e"]IT entails "A believes that B will do e" or a similar psychological fact about A. Even if this inference is so, the strength or weakness of A's conviction concerning the doing of e by B cannot in itself lend any credibility to the proposition that B must do e or that B does not do e freely. We do not judge an action unfree or necessitated merely because we find someone who believed, felt certain, or was absolutely convinced that the action would be done.

But, it may be replied, knowing is a different state from feeling certain or being absolutely convinced. For knowledge is the only state that guarantees the occurrence of the event foreknown. Thus if Jones at t really is in the state of knowing that Smith will do e at t + n, then it is logically necessary for Smith to do e at t + n.

Thus, one has the picture of a state of the knower at one time mysteriously necessitating or constraining some future event. But this is a confusion that is easily cleared up by looking more closely at the term "state."

Some descriptions of the state of an object are true only if certain future events occur. Citation is the winner of the horse race only if the judges later certify that he carried the correct weight during the race. No one believes that this fact lends credence to the belief that some state of Citation's prior to the judges' decision constrained them to decide that Citation carried the correct weight. "Knowing" may be like "winning" insofar as the non-occurrence of the event alleged to be known entails that there was no knowledge. But this certainly does not warrant the conclusion that the knowing brings about or constrains the event foreknown just as Citation's winning does not bring about or constrain the judges' decision. It does show that "knowing" is not like "being in pain" insofar as my now being in pain entails nothing about the future.

Hence the belief that foreknowledge and freedom are incompatible is not supported by consideration of psychological facts about the knower.

If one feels that the basis of the inference from "A knows that B will do e" to "B must do e" or "B does not do e freely" lies in the fact that "A knows that p" entails "p," then he is mistaken. There is, it is true, a relationship of logical necessity between the two propositions "A knows that B will do e" and "B will do e" the former entailing the latter. Clearly, though, to infer validly that B's action is not free, one would have to show that the act is necessary or that B must do e in a different way. One cannot say that B's act is necessary in a sense of "necessary" which precludes freedom merely because the sentence describing the act follows necessarily from the true sentence "A knows that B will do e," for there is always a true sentence which entails a sentence describing any particular event. If "Jones washes his car on May 22, 1984" is true, it follows from the true sentence "Jones washes his car on May 22, 1984 -2 + 2 = 4" or simply "Jones washes his car on May 22, 1984." But it surely does not follow that Jones does not wash his car freely on May 22, 1984. I am not here denying that foreknowledge and freedom are incompatible. If they are, however, it is not because of the logical necessity which holds between "A knows that p" and "p." We would have to deal with aspects of the content of the knowledge claim other than this one. If A truly asserts today: "Jones washes his car on May 22, 1984 · 2 + 2 = 4," then, if we assume that A has absolutely no reason to believe the first conjunct, we are not in the least inclined to feel that Jones had to wash his car; whereas if A truly asserts today: "I know that B will do e tomorrow," we may very well be inclined to view B's action as necessary. Thus, some feature of the knowledge-claim must be responsible for our inclination, other than its being true and entailing the sentence describing the event whose occurrence A claims to know.

Many philosophers would assert that knowledge-claims imply the ability to support the claims by reliable evidence or good reasons. There are three major sources of disagreement.

First of all, some have suggested that there are propositions which we know to be true for which the request for evidence is inappropriate. But the propositions in question are not propositions of the kind "B will do e." They are either logico-mathematical propositions or propositions which are known "directly," e.g., "I am in pain now," "It seems to me that there is a red object before me now," and the like.

Miss Anscombe speaks of "knowledge without observation." For example, I ordinarily know without observation that my limbs are in a certain position or that I have kicked my leg. Let us grant that these cases are correctly described as cases of "knowledge without observation." Let us also grant that non-observational knowledge is knowledge concerning which the request for evidence is inappropriate. But propositions of the form "A knows that B will do e" cannot be cases of non-observational knowledge if A is a different person from B since non-observational knowledge is always self-knowledge. I cannot know without observation the position of your limbs. If, on the other hand, A is the same person as B, there are two possibilities: (1) A may know that he will do e in the same way other people know that he will do e. Although these kinds of cases have been underemphasized in recent literature, they do exist. To the question, "How do you know that you will insult him?" A may reply: "People like that always irk me." But, of course, this kind of knowledge is not non-observational or, at least, not nonevidential. A is appealing to the same kind of evidence that other people must appeal to. (2) A may know that he will do e in a way which precludes the necessity of his providing evidence. For example, if A asserts "I shall wash the car tomorrow," it may be reasonable to infer that he will wash the car tomorrow (although it would be odd for A to say that he knows that he will wash the car tomorrow unless someone suggests to him that there is reason to believe that he will not). It is interesting that, when A does wash the car tomorrow, we are not ordinarily led to feel that he could not have done otherwise. If A's utterance was the expression of a decision or an intention, then we may feel that he could have changed his mind. Moreover, libertarians sometimes contend that decisions, rather than actions, are the proper subjects of the predicates "free" and "unfree." Hence, from the information that A has made a certain decision and has implemented that decision, we are not led to feel that he lacked freedom. When certain information is brought to bear, however, we may begin to feel that A could not have done (or decided to do) otherwise. As we shall see shortly, this information comprises the evidence that would support another person's knowledge claim about the doing of e by A. Thus, even if there is non-observational knowledge, and even if one can say of a person that he can have non-observational knowledge of his own future actions, the cases that induce concern about a man's freedom are those in which knowledge of future action is supported by evidence, either from the person acting or some other person. A second center of disagreement with the assertion that knowledge-claims require evidential support is among those who view knowledge as a special state of mind vis-a-vis the object of knowledge. The untenability of this doctrine has been demonstrated by a number of philosophers, e.g., A. J. Ayer, and it is for this reason that I dismiss it from consideration.

Finally, some philosophers have replaced this condition, viz., the ability to support the knowledge-claim, with the presence of some sort of causal relation between either the known fact and the belief, or certain facts related in a specified way to the known fact and the belief. In the case of "A knows that B will do e," the facts that cause A's belief cannot include "B will do e" for this is a future fact. But the present state of affairs that causes B to do e in the future can cause A's belief, and if so, A's belief is knowledge. Generally, philosophers who support this causal theory of knowledge point out that all the conditions of knowledge, including the causal condition, can be satisfied even where the knower cannot specify the grounds of his knowledge.

Let us first consider the consequences of the assumption that "A knows that B will do e" is true only if A has reliable evidence to warrant the proposition "B will do e." Suppose A is a psychologist who has excellent reason to believe that "AU people with characteristics a, b, and c, when placed in circumstances s, will do e" is a law; further, suppose he asserts that he knows B will do e on the basis of his knowledge that B satisfies the antecedent of this presumed law. Now, if we are inclined to accept the inference from "A knows that B will do e" to "B does not do e freely" as valid, I suggest that our inclination is due to the existence of such excellent grounds for the proposition "B will do e." We have seen that the inference cannot be supported via a consideration of the other alleged features of knowledge-claims. Its support must lie in the necessity of offering evidence to support knowledge-claims. Consider a knowledge-claim put forth by A, e.g., "I know that B will choose steak at the restaurant tonight." Suppose that A is convinced that B will do this, and suppose that B does do it. We also know that A's belief is not based on information about B's eating habits or on any reliable evidence whatsoever. A is asked: "How did you know that B would choose steak?" He replies: "I just felt he would." We may then say to A:" You didn't know. You just guessed." The point to be stressed is that we do not view B's action in this situation as unfree or necessitated since A's knowledge-claim could not be supported by information about B. Suppose, on the other hand, A replied: "I have been doing a detailed psychological and physiological investigation of food choices and I have found that they depend on six factors. B, of course, was not informed of these results, in order to avoid the introduction of extraneous factors. I studied B and his situation, and then calculated that his choice would be steak." We might consider that poor B could not have chosen otherwise, and our feeling would clearly be due to the existence of such excellent evidence.