Unknowability of God In Jewish Rationalism II: Not Negative Theology

Note: I am trying to work some mathematical notation into these posts, gratuitously at first, but later with the purpose of formalizing ideas. If you use the Firefox browser, you should be able to see the math no problem, I hope. If you use Internet Explorer, you may need to download the free MathPlayer plug-in from Design Science. (Because Microsoft makes things easy.) If you can see the red equation below (rather than a bunch of text), you should be OK:

`sum_(k=1)^n k = 1+2+ cdots +n=(n(n+1))/2`

OK. On with the show...

The greatest Jewish medieval rationalist proponent of the unknowable God is without any doubt Maimonides. His approach is frequently known as the "negative theology" or the doctrine of "negative attributes." The basic notion of negative theology probably dates back to Aristotle or Plotinus, but our interest here is not in tracing the intellectual history of the idea, and so we simply begin with Rambam. On a superficial level, the idea behind negative attribution is that instead of affirming attributes of God (by saying, for example, that God possesses "will" or "power"), it is appropriate that one should deny their opposites or privations (by saying, for example, that God is not lacking "will" or "power"). A typical selection from Rambam is as follows (GP I:58):

... we say of Him [God], because of these notions, that He is powerful and knowing and willing. The intention in ascribing these attributes to Him is to signify that He is neither powerless nor ignorant nor inattentive nor negligent... every attribute that we predicate of Him is an attribute of action or, if the attribute is intended for the apprehension of His essence and not of His action, it signifies the negation of the privation of the attribute in question... you come nearer to the apprehension of Him, may He be exalted, with every increase in the negations regarding Him...

The question that immediately leaps to mind is whether and how any advantage can possibly be gained by merely replacing affirmations with negations. Basic acquaintance with logic suggests that this approach will not achieve anything at all, and this was appreciated by all the significant medieval thinkers. Duns Scotus, for example, writes

For every denial is intelligible only in terms of some affirmation. It is also clear that we can know negations of God only by means of affirmations; for if we deny anything of God, it is because we wish to do away with something inconsistent with what we have already affirmed.

Crescas, I believe, makes the same point, as does Albo (husik_29):

For although it is true that all negations are predicable of God, still no wise man can negate any attribute unless he knows how the positive attribute applies to the thing characterized by it, and understands the aspect of perfection as well as of defect which the attribute contains.

It is clear that the negation of an attribute cannot be meaningful without the possibility of its affirmation, and that for classical logic the double-negation must equal the affirmation, `¬(¬x)=x`. In short, the problem is that without positive attributes there cannot be negative attributes (berkovits_04). Rambam himself was equally aware of the obvious difficulty in this interpretation of the negative attributes, and although he does give what appears to be a defense of this very position in GP I:60, he takes greater pains to explain that this superficial negation is not the sort of negation he is advocating. Rather, the negation that Rambam ultimately advocates is the negation of the very applicability of any attribute to God. Thus (GP I:58),

... those negations are not used with reference to or applied to Him, may He be exalted, except from the following point of view, which you know: one sometimes denies with reference to a thing something that cannot fittingly exist in it. Thus we say of a wall that it is not endowed with sight.

Thus, the application of any attribute to God constitutes what we today sometimes call a "category error." Just as there is some set of attributes which do not apply to walls (e.g., sight, wisdom, gender, etc.) and which can therefore not meaningfully be affirmed or denied of walls, there is a set of attributes which do not apply to God and which can therefore not meaningfully be affirmed or denied of God. Thus, Rambam's negation means that there are no attributes whatsoever that may be predicated in the same sense (i.e., univocally) of both humans and of God, a position which may be referred to as the doctrine of "absolute equivocation." More simply, as Steinsaltz puts it, we have no terminology to describe God. Thus, the predication of any attribute in regard to God is a predication which can neither be affirmed nor denied, but is one which falls into a critical third logical class of which the classical logicians often failed to take note:

A proposition must be either true or false, provided that the predicate be one which can in any intelligible sense be attributed to the subject (and as this is always assumed to be the case in treatises on logic, the axiom is always laid down there as of absolute truth). "Abracadabra is a second intention" is neither true nor false. Between the true and the false there is a third possibility, the Unmeaning ..." (mill_36)

Since Mill's time there has been some work on devising logics which incorporate the "unmeaning" valuation of predicates (see Haack, 1974), but that would be getting far afield for us, as it is hardly what Rambam could have had in mind. On Rambam's view, in the case of God as the "subject," the set of "Unmeaning" propositions is just the set containing every proposition. Let us then look at Rambam's position in his own words:

Now everything that can be ascribed to God, may He be exalted, differs in every respect from our attributes, so that no definition can comprehend the one thing and the other. (I:35)

... it behooves those who believe that there are essential attributes that may be predicated of the Creator—namely that He is existent, living, possessing power, knowing, and willing—to understand that these notions are not ascribed to Him and to us in the same sense... clear it is to all those who understand the meaning of being alike that the term "existent" is predicated of Him, may He be exalted, and of everything that is other than He, in a purely equivocal sense. Similarly, the terms "knowledge," "power," "will," and "life," as applied to Him, may He be exalted, and to all those possessing knowledge, power, will, and life, are purely equivocal, so that their meaning when they are predicated of Him is in no way like their meaning in other applications. (I:56)

Seeskin explains Maimonidean absolute equivocation in this way:

In God's case, the denial not only rejects the predicate in question but puts God outside the scope of the predicate or anything like it. To the question "How do we know which negations to make?" Maimonides' answer is that at one level we negate predicates that suggest imperfection to the average worshiper, but at another level we negate all predicates. In either case, there is no common core of meaning once we take into account the full extent of the negation (seeskin_00).

The only thing we can say about the nature of God is that whichever way we view it, or in whatever way we describe it, it is off the scale of intelligibility. In one way, this admission says more about us than it does about God. It says that when we try to understand God, we come face to face with our own limitations... (seeskin_91)

The negative theology is also strongly advocated by Albo, whose method he sees foretold in the "still small voice" of I Kings 19:12 (husik_29, 217). Concerning, in particular, the understanding of God's volition, he writes:

For God's will follows His wisdom and His power, and as His wisdom and His power are infinite and not of the same kind as our wisdom and power, so His will is not of the same kind as our will. And as the wisdom and knowledge of God are not something added to His essence, as we shall see later, but are of His very essence, and His essence is absolutely unknown, His wisdom is also absolutely unknown. And when we attribute wisdom to God, all that we understand is that we deny the attribute of ignorance, which is the opposite. In the same way we attribute will to God in the sense that we deny that He acts by necessity like the defective agents of finite power, God's activity being with infinite power. (husik_29, 20)

The reason why, on Albo's view, we even bother making attributions at all, is simply for the sake of appearances: God cannot be considered inferior to any merely physical contingent being, and since, for example, "volition" is a perfection among physical beings, we simply cannot deny that God has volition. Similarly, "We ascribe knowledge to Him because it is a perfection which it is inconceivable that God should be without." Fundamentally, it would be unseemly to deny to God attributes that are considered perfections among his creations. Still, Albo does not uniformly deny the applicability of all attributes, but distinguishes between desirable attributes (perfections) and undesirable attributes (imperfections). Only the former need be denied by way of absolute equivocation, while the latter can be denied in their mundane sense:

...not all attributes are negated in the same manner... If the attribute denotes a defect, it is negated in another way, namely in respect to its defective element. To deny everything of God without reason or understanding would not indicate God's perfection. All those attributes which we regard as defects, like dead, weak, ignorant, poor, bad, and so on, we negate absolutely, meaning that such attributes, which are defects, are not found in God, but that the opposite, which is a perfection, is found in Him. But when we negate the other opposite, which denotes a perfection in our estimation, like living, powerful, wise, rich, good, the meaning is not that this attribute does not exist in Him but the opposite, which is a defect, does. Such negation, far from ascribing perfection to God, is insult and blasphemy. What we do mean in the latter case is that the perfections which we understand by the terms living, powerful, wise, and so on, are not found in God as they are found in us but in a manner more honorable and more excellent, so much so that there is no relation between the perfection in us or as we understand it and the perfection when ascribed to God. The perfection or perfections in question are applied to man and to God as homonyms, considering the enormous difference that there is between them. (husik_29, 200)

I don't find Albo's remarks here to be particularly coherent, and perhaps his conclusion that "the most that we can understand about God is that we can not understand Him" (husik_29, 206) is an indication that Albo is not entirely pleased with his own analysis either. Returning to Rambam's "way of negation," or the doctrine of absolute equivocation, while one can justifiably admire Maimonides for this bold stroke of genius, it quickly becomes apparent that nothing was solved.

It is clear what price Maimonides pays for rejecting the doctrine of analogy. If there is no analogy between God's intelligence, power, and goodness and ours, we have no hope of understanding what these attributes are in God. We can say that they all must be identical or else God's unity would be compromised. But when it comes to how God thinks, how He generates power as a result of thinking, and why His power is necessarily benevolent, the only thing we know is that we cannot know. All we can do is confess ignorance. This is another way of saying that the internal nature of God is and will always remain a mystery... (seeskin_91)

The problem here is that while it is fine on the one hand to declare that predicates applied to God do not have the same meaning as predicates of the same name applied to man (i.e., that all such predicates are applied absolutely equivocally), the fact remains that all of the crucial arguments that the medieval rationalists offer in favor of God's existence, knowledge, power, incorporeality, etc., rely on the very fact that these two sets of predicates do mean the same things in both cases! This critique is clearly formulated by Gersonides in Wars of the Lord:

...since it is clear when we deny attributes of God that are found in us that such attributes are not completely equivocal with respect to God (may he be blessed) and us, the same is true when we affirm of God predicates that are true of us. For example, we say that God is immovable, since if He were movable He would be a body, for all moveable objects [are bodies]. Now it is evident that in this proposition the term "moveable" is not completely equivocal with respect to the term "moveable" when it is applied to nondivine things. For if it were, there would be no proof that God is not moveable, since the moveable object that must be a body is that which is moveable in the domain of human phenomena, whereas the term "moveable" (in the completely equivocal sense) would not imply that it is a body. Hence, since it is evident that the predicates we deny of God are not absolutely equivocal, neither are the terms that we affirm of Him. In general, if the terms used in affirming predicates of Him were absolutely equivocal, there would be no term applicable to things in our world that would be more appropriate to deny than to affirm of God or [more appropriate] to affirm that to deny of Him. For example, someone could say "God is a body" but not mean by the term "body," "a magnitude"; rather he would mean something that is completely equivocal with the term "body" as we usually use it. Similarly, someone could say "God does not have knowledge," since the term knowledge would not [on this view] have the same meaning for him in this statement as it does for us. (v.2 p.110, Feldman)

Ralbag's argument basically says that if an attribute is predicated of God in a completely different (and unknown) sense than that in which it is predicated of man, then when such a predicate is applied to God, it is no longer possible to use that term in any aspect of theological reasoning. Samuelson (1969) puts it succinctly:

Thus, in the case of a system of God-talk based on Maimonides' account, it makes no more sense to say, for example, "God is good" than to say that "xtqpr is qpryiz," since in both cases there are no rules governing the use of the constituent terms of the proposition. Actually, it is not even correct to call either letter sequence a proposition, since neither sequence constitutes anything that is in any way intelligible. Both are plain and simple gibberish.

The argument of Gersonides seems to me to be entirely cogent and final, although I might not go so far as Samuelson to say that "this argument is devastating against any form of the position that God is beyond human comprehension." It is only fatal against the use of inferential procedures involving the attributes of God, not against His essential unknowability. In any case, it is at least in part because of this thorny issue that Maimonides' position was rejected by most subsequent philosophers (wolfson_53). As to why his views were rejected among the masses, the greater problem may have been due to the inadequacy for religion of a propertyless, attributeless, featureless God, even if it could be shown conclusively that this is the only logically coherent understanding of God available within a radical monotheistic framework. As Berkovits puts it, "The negative attributes will never do. Religion cannot forgo the love and the mercy of God, or even his justice and anger. Such attributes have to be related to him in a positive sense, or else there is no basis for a living God or religious observance."

However, despite the dramatic distance that Rambam puts between man and God, he still strives to limit this barrier only to matters pertaining to God's essence or quiddity (God's "what-ness"). He seems to reserve room for knowledge of God in areas that do not impact the question of His quiddity, such as those areas that relate only to God's actions. We will try to assess this distinction below, but first we will make a short detour into the "problem of predicates," in which we try to gain some insight into why Rambam and other medieval thinkers were so adamant about the "indefinability" of God. Why can we not have a definition? Why can attributes only be applied to God equivocally? What would be so loathsome in attributing to God "knowledge" or "will" or "agency" in the same sense that we ourselves possess these attributes? Clearly, the medieval thinkers saw grave dangers in pursuing such a course. What was it they saw?