20th World Congress of Philosophy Logo

Medieval Philosophy

Al-Ghazâlî, Causality, and Knowledge

Peter Adamson
University of Notre Dame

bluered.gif (1041 bytes)

ABSTRACT: Few passages in Arabic philosophy have attracted as much attention as al-Ghazâlî's discussion of causality in the seventeenth discussion of Tahâfut al-Falsafa, along with the response of Ibn Rushd (Averroës) in his Tahâfut al-Tahâfut. A question often asked is to what extent al-Ghazâlî can be called an occasionalist; that is, whether he follows other Kalâm thinkers in restricting causal agency to God alone. What has not been thoroughly addressed in previous studies is a question which al-Ghazâlî and Ibn Rushd both see as decisive in the seventeenth discussion: what theory of causality is sufficient to explain human knowledge? In this paper I show that al-Ghazâlî's and Ibn Rushd's theories of causality are closely related to their epistemologies. The difference between the two thinkers can be briefly summerized as follows. For Ibn Rushd, the paradigm of human knowledge is demonstrative science; for al-Ghazâlî, in contrast, the paradigm of human knowledge is (or at least includes) revelation. Yet both remain committed to the possibility of Aristotelian science and its underlying principles. Thus, I suggest that al-Ghazâlî's stance in the seventeenth discussion sheds light on his critique of philosophy in the Tahâfut: namely, philosophy is not inherently incoherent, but simply limited in scope. I also briefly compare this position to that of Thomas Aquinas, in order to place the view in a more familiar context.

bluered.gif (1041 bytes)

Few passages in Arabic philosophy have attracted as much attention as al-Ghazâlî's discussion of causality in the seventeenth discussion of Tahâfut al-Falâsifa, along with the response of Ibn Rushd (Averroës) in his Tahâfut al-Tahâfut. A question which has been addressed several times is to what extent al-Ghazâlî can be called an occasionalist: whether he here follows other Kalâm thinkers in restricting causal agency to God alone. (1) What has not been addressed in studies of this text is a question which al-Ghazâlî and Ibn Rushd both see as decisive in the seventeenth discussion: what theory of causality is sufficient to explain human knowledge? (2) In this paper I will show that al-Ghazâlî's and Ibn Rushd's theories of causality are closely related to their epistemologies. The difference between the two thinkers can be quickly summed up by saying that for Ibn Rushd the paradigm of human knowledge is demonstrative science, whereas for al-Ghazâlî the paradigm of human knowledge is, or at least includes, revelation. In closing I will suggest that al-Ghazâlî's commitment to this paradigm sheds light on the guiding intent of his critique of philosophy in the Tahâfut.

But before turning to the epistemic aspect of the seventeenth discussion, let me say briefly what I take to be al-Ghazâlî's basic position on causality. As others have noted, his critique here imputes a very strong notion of causality to the philosophers: namely that given the existence of a cause, the existence of its effect is necessary. (3) Al-Ghazâlî holds that, on such a notion of causality, only God is a cause. This is because, given the existence of miracles, and accepting the proposition that God can do anything, no cause other than God can necessitate its effect. It is always possible that God might will the expected effect not to proceed, or will an entirely different effect to proceed. Al-Ghazâlî defends this view against both philosophers who claim that a natural cause, such as the fire which causes the burning of cotton, is the sole and sufficient cause for its effect, and against those who, like Ibn Sînâ (Avicenna), would say that there is a giver of forms in the celestial world which imposes form once a sublunar cause has prepared some matter for that form. Against the first view, al-Ghazâlî gives the famous argument which has been compared to Hume's: observation of simultaneity does not prove that causation has occurred. Against the latter view, al-Ghazâlî says that if effects are brought about by higher principles, they depend ultimately on God's will, and God can do anything except the absolutely impossible. Therefore, no effect proceeds necessarily from its cause, unless the cause in question is God Himself.

But al-Ghazâlî goes on to say, in essence, that natural causes can be regarded as causes if we invoke a weaker notion of causality. He admits that a natural cause has a nature which gives rise to certain effects: fire, for instance, has a nature such that it burns whatever is in contact with it. But this does not mean that fire is a necessary cause, in the sense that the existence of fire in contact with cotton logically entails the existence of burning cotton. The nature of fire itself, says al-Ghazâlî, derives from God, and God chooses whether or not this nature will give rise to its normal effect or not. On al-Ghazâlî's view, natural causes are only contingently causes -- their effects only proceed if the true Agent who gave them their natures wishes it. (4) Ibn Rushd was the first of many to see this position as an inconsistent concession to the philosophers on al-Ghazâlî's part, because al-Ghazâlî seems at first to say that God is the only cause, and then asserts that created things do have natures which lead them to cause their effects. But al-Ghazâlî's view is not inconsistent: it merely sees created natures as inherently contingent and provisional, relying on God's continued will for their efficacy and very existence.

Ibn Rushd's response to al-Ghazâlî is based on the objection that if any form of occasionalism is accepted, there is no possibility for human knowledge. He remarks that if what he perceives to be al-Ghazâlî's denial of causality is accepted, 'there is no fixed (thâbit) knowledge of anything,' because 'certain (yaqînî) knowledge is the knowledge of the thing according to what it is in itself.' (5) Here Ibn Rushd is tacitly relying on two principles of Aristotelianism. The first is what we might call epistemic optimism: as an Aristotelian, Ibn Rushd takes it for granted that humans have knowledge, so that if a theory of causality is incompatible with our knowing things as they are, then this is itself a sufficient refutation of that theory. The second is the principle that things are only known demonstratively through their causes. This is of course a basic tenet of medieval epistemology: elsewhere in Arabic philosophy, it is used to argue for the impossibility of positive theology in the Liber de Causis, and it is the basis for St. Thomas Aquinas' notion of a propter quid demonstration. (6) For Ibn Rushd these two principles are the guidelines along which a theory of causality must be developed. He says, for instance, that 'if the things whose causes are not perceived are unknown by nature and sought after, then what is not unknown necessarily has perceived causes.' (7) On the strength of the two principles, he is confident that he can rule out any form of occasionalism.

The first thing that should be noted about Ibn Rushd's objection to al-Ghazâlî here is that it does not seem, at first, to be relevant. The objection assumes that al-Ghazâlî rejects the idea that things have natures, which are both the principles of demonstrative knowledge and the principles of causality. By denying causality, Ibn Rushd argues, al-Ghazâlî has also rejected the possibility of knowledge. But as we saw above, al-Ghazâlî does not in fact completely reject natures: he thinks a created thing has a created nature which causes a proper effect, but that this nature and causation are always subject to God's will. Still, one can readily see how Ibn Rushd's objection might be adapted to meet this rather less occasionalistic view. If natures only possibly give rise to their effects, then our knowledge of them is not necessary, but only probable: natures must remain always the same if they are to be the objects of demonstrative, scientific knowledge. (8) Ibn Rushd alludes to this requirement when he remarks that knowledge based on natures must be 'fixed' -- in other words, it must always be the case that natures cause a proper effect, by the very definition of 'nature.' (9) This is one reason why Ibn Rushd insists that miracles, i.e. those cases in which God does step in and disturb the natural course of causation, are not even something we can rationally discuss. Supernatural events, as the very name implies, are beyond any knowledge we can have of nature, and therefore are not properly to be included in any discussion of causality or philosophy in general.

It is a measure of how inapt is the comparison so often made between al-Ghazâlî and Hume that al-Ghazâlî both anticipates this objection and takes it seriously. His goal is certainly not a kind of Humean skepticism; on the contrary, he is just as committed as Ibn Rushd to Aristotelian epistemic optimism, and in a sense he is even committed to the principle that knowledge is only through causes. Thus al-Ghazâlî himself raises the question of why miracles do not prevent our knowledge of the empirical world, admitting that if they did, a man who left a book in his home would have to say, 'I do not know what is in the house now, and the extent of what I know is only that I left a book in the house, and perhaps now it is a horse.' (10) Al-Ghazâlî's response to the objection is most intriguing: he suggests that God continually creates in us the knowledge that He will not perform these miracles. Thus the source of a man's knowledge that, say, his book is still in the house, is God Himself. Indeed, al-Ghazâlî seems to be contrasting the so-called 'knowledge' of experience, which only leads to the habit of expecting given natures to cause given effects, with a certain knowledge created in us by God. It should be noted here that it has been questioned whether this passage represents al-Ghazâlî's own views -- the dispute turns on a point of translation. (11) We need not, however, decide this question here, because even outside this passage al-Ghazâlî continues to propose that knowledge can be created in us by God, both in the seventeenth discussion and elsewhere. (12)

A modern reader is likely, I think, to reject al-Ghazâlî's solution as inadequate: a knowledge created by God in a human hardly seems to qualify as knowledge in the proper sense at all. This is in fact exactly the objection put by Ibn Rushd, who argues that one only has knowledge if that knowledge has a direct relationship to the natural cause that is known. Even assuming that God does create knowledge in a person, that person is only said to know if the knowledge 'is something dependent on the nature of the existent, because the true is [when] one believes something to be as it is in existence.' (13) Thus our two authors are, at root, at odds with one another over the epistemic question: what are the conditions for certain knowledge? For al-Ghazâlî, the habit of knowing brought about by experience is not knowledge of what is necessary; only a knowledge produced by God is certain. For Ibn Rushd, the situation is precisely reversed: if God creates a knowledge in us, that knowledge is properly designated as knowledge only if it corresponds to a real nature.

Modern intuitions about this question notwithstanding, it must be admitted that al-Ghazâlî's view on justification is perfectly consistent with his view on causality. Al-Ghazâlî admits that we do know that certain things will and will not happen. But if causes are always contingent on God's will in producing their effects, then certain knowledge can only derive from the real source of necessity in the causal relationship, namely God. Thus, as already remarked, there is a sense in which al-Ghazâlî would agree to the second Aristotelian principle that knowledge is always through causes. But for him this means that certain knowledge is always through God, because causality is only through God. That this is really al-Ghazâlî's intent can be seen from the fact that he puts knowledge of normal events on the same plane as the special knowledge that is enjoyed by prophets. The knowledge that comes about through the habitual course of nature is just as certain as the knowledge of a prophet that there will be an exception in the course of nature, because both sorts of knowledge are created by God. (14) Thus it makes sense to say that, for al-Ghazâlî, revelation or prophesy is the paradigmatic form of knowledge for humans.

In the face of this view, Ibn Rushd's insistence on the Aristotelian definition of knowledge as perception in conformity with a nature seems to be little more than begging the question. This is because, as already mentioned, Ibn Rushd assumes that the supernatural -- which for al-Ghazâlî is fundamental -- is a subject which cannot be understood nor even rationally discussed. Indeed, he chastises al-Ghazâlî repeatedly for bringing into the arguments of the Tahâfut things that should not be disputed, lest the political function of religion be compromised. In the seventeenth discussion, Ibn Rushd's political argument against al-Ghazâlî is buttressed by an epistemological one: knowledge as such has to do with natures, and therefore excludes supernatural causes and events. Whether Ibn Rushd believes that miracles really do happen, or are only said to happen for the sake of common believers, they will by definition not fall within the realm of scientific, Aristotelian discourse: 'it is necessary to say about [the shari'a] that its principles are divine matters transcending human intellects, and they must be recognized without fail despite ignorance of their causes.' (15) By itself, this assertion is question-begging because it does not address possible counter-examples to the 'philosophical' paradigm of knowledge -- such as miracles -- assuming in effect that 'cause' and 'knowledge' are simply terms which do not apply to the supernatural.

Ibn Rushd does, however, have a more elaborate response to give against the accusation of question-begging, because of his own account of the importance of God for natural causes. The regularity and predictability of natural causes, according to Ibn Rushd, is a testament to God's wisdom. Ibn Rushd can gain support for this view from the Qurân, which he quotes explicitly: 'And you will not find any alteration in the doings (sunna) of God, and you will not find any change in the doings of God.' (16) Scientific knowledge is not, then, an affront to God's power, but is only made possible by God's wisdom, which does not allow natures and therefore causal relationships to change. The defense is deepened by Ibn Rushd's allusion to another doctrine he holds in the Tahâfut, namely that God's knowledge of things is the cause of their existence. Ibn Rushd goes so far as to remark that God's knowledge of natures must have the same objects as our own, though in God's case the relationship is causal, as well as epistemic: 'if we have knowledge of these possibles, then there is a condition (hâl) in the possible existents to which our knowledge pertains... and this is what the philosophers designate as nature. Likewise, the knowledge of God is through the existents, although [God's knowledge] is their cause... and therefore it is necessary that the existents come about in accordance with His knowledge.' (17) But the rejoinder open to al-Ghazâlî is clear: Ibn Rushd has here effectively raised scientific knowledge to the level of divine knowledge; indeed, he has identified the two. Thus it also becomes clear how much the paradigm of knowledge, for Ibn Rushd, is that of natural science. But against al-Ghazâlî, the view remains question-begging. This is because Ibn Rushd assumes, rather than argues, that divine wisdom and knowledge would be incompatible with a change in the course of nature, whereas al-Ghazâlî insists precisely that such changes are possible.

Yet, does not Ibn Rushd have a legitimate criticism of al-Ghazâlî insofar as al-Ghazâlî's stance does not allow for scientific knowledge at all? If al-Ghazâlî holds to some kind of epistemic optimism regarding natural science, this should remain a problem for him. The answer to this question sheds light on al-Ghazâlî's general attitude towards philosophy in the Tahâfut. Recall that al-Ghazâlî does, in fact, concede to the philosophers that there are natures which give rise to effects. So he leaves room for scientific knowledge of those natures (e.g. the knowledge that fire burns). What al-Ghazâlî denies is that such knowledge constitutes necessary knowledge -- scientific discourse is partial, because it cannot establish whether a given natural cause will be superseded by supernatural intervention. Thus al-Ghazâlî does not reject scientific or philosophical knowledge altogether. What he does do is to show that it does not measure up to the rather high standard the philosophers have set for themselves, namely that knowledge be of relationships which are logically necessary.

This is, I would submit, representative of al-Ghazâlî's strategy in the Tahâfut. His goal is not to show the 'incoherence' of the philosophers, if this is taken to mean that all of philosophy is incoherent. His goal is rather to show that philosophy must be subsumed within an intellectual enterprise which includes revelation -- and further, that revelation must be regarded as the most superior kind of knowledge. The strategy of the Tahâfut is to show that philosophy, mostly as represented by Ibn Sînâ, has overreached itself in a number of cases, reaching false conclusions where reason should simply be recognized as inadequate. Thus in the seventeenth discussion he remarks, regarding prophetic miracles, that 'although the extent of [the prophet's] power is not determined in the intellect, there is still no need to deny what is traditionally handed down and what revelation mentions.' (18) Similarly, in the second discussion, taking an approach like that of Maimonides, al-Ghazâlî argues, 'since it is clear that we do not at all dismiss the permanence of the world from the viewpoint of the intellect, but allow as possible the permanence and extinction [of the world], we only know which possibility is really the case through revelation; thus insight into this does not belong to reason.' (19)

Although al-Ghazâlî has traditionally been seen as an anti-rationalist, a mere opponent of philosophy, our analysis of the seventeenth discussion suggests that his attitude towards philosophy is both more subtle and less radical. The al-Ghazâlî of the Tahâfut should be classed not with the anti-rationalists, but with those concerned to keep philosophy in its proper place. For an analogous position in medieval Europe, we need look no further than St. Thomas Aquinas. Though Aquinas is usually considered to be far more of a rationalist than al-Ghazâlî, a Ghazâlîan strategy of limiting the claims of philosophy can be found in, for example, Question 12 of the Prima Pars, where Aquinas argues that human reason by itself is unable to know God, so that humans require the supernatural assistance of grace to achieve their perfection. This is not to suggest that Aquinas' view on the more specific issue of causality is comparable to al-Ghazâlî's. Rather, it is to suggest that, like Thomas, al-Ghazâlî can be seen as sympathetic to philosophy within its proper limits, even though he holds that philosophy is not the highest paradigm of human knowledge.

bluered.gif (1041 bytes)


(1) An argument against the occasionalist reading of al-Ghazâlî is mounted in L.E. Goodman, "Did al-Ghazâlî Deny Causality?," Studia Islamica 47 (1978), pp. 83-120. A more occasionalist reading is given in Binyamin Abrahamov, "Al-Ghazâlî's Theory of Causality," Studia Islamica 67 (1988), pp. 75-98, which also takes account of al-Ghazâlî's writings outside the Tahâfut. Ilai Alon, "Al-Ghazâlî on Causality," American Oriental Society Journal 100 (1980), pp. 397-405, argues that al-Ghazâlî is trying to compromise between the theological and philosophical views. Another discussion can be found in Stephen Riker, "Al-Ghazâlî on Necessary Causality," The Monist 79 (1996), pp. 315-324.

(2) The only treatment of this issue I have found is Michael Marmura, "Ghazali and Demonstrative Science," Journal of the History of Philosophy 3 (1965), pp. 183-204. Marmura raises some of the same topics I treat here, notably the tension between knowledge and occasionalism and the importance of the creation of knowledge by God. Marmura's reading of al-Ghazâlî is, however, more occasionalist than my own, so that we reach different conclusions about al-Ghazâlî's epistemology.

(3) See Goodman, pp. 84-5.

(4) Al-Ghazâlî's view is reminiscent of the one outlined by al-Kindî in his short treatise On the True Agent, in Al-Kindî, fî al-Falsafa al-Ûlâ, in Risâ'il al-Kindî al-Falsafiyya, ed. Abû Rîda (Cairo 1950), vol. 1, pp. 182-4, English translation in A. Altmann and S.M. Stern, Isaac Israeli (Westport, 1979), p. 68-9. In this treatise al-Kindî argues that though God is the true Agent, His effects are agents in a secondary, metaphorical sense, through their dependence on Him.

(5) Tahâfut al-Tahâfut li-Ibn Rushd, ed. Muhammad al-'Arîn, Dâr al-Fikhr al-Lunbânî (Bairût, 1993), p. 296. (Hereafter Tahâfut.) English translation: Averroes' Tahâfut Al-Tahâfut, trans. Simon Van Den Burgh, EJW Gibb Memorial Trust (Cambridge, 1954), p. 325. I read yaqînî with Bouyges and al-'Arîn, rejecting Van Den Burgh's reading of haqîqî ('true'). In the notes to his translation, Van Den Burgh comments that he takes this to be an adequate refutation of al-Ghazâlî (Van Den Burgh, vol. 2, p. 177). All translations in the paper are my own, though I have consulted Van Den Burgh's translation and given page citations to it in the foonotes.

(6) Liber de Causis, Prop. 5: 'And each thing is only known and described from its cause. And when the thing is only a cause and not an effect, it is not known by a prior cause, and is not described because it is higher than description, and speech does not reach it.' (I translate from the Arabic.)

(7) Tahâfut, pp. 290-1. English translation p. 318.

(8) A similar point is made by Marmura, p. 200.

(9) Of course, this does not rule out that something else may impede a cause from bringing about its effect, as in the case where something is in contact fire but does not burn because it is covered with talc. Ibn Rushd addresses this point, remarking that the causation may be hindered (and is in this sense not necessary), but this does not mean fire has lost its 'name and measure (hadd),' i.e. the nature which makes it cause burning. (Tahâfut, p. 291. English translation p. 319.)

(10) Tahâfut, p.295. English translation p. 324.

(11) Goodman (p. 105) points out that the Van Den Burgh translation omits the words 'fa laysa fî hâdhâl-kalâm illâ tashnî' mahd' meaning 'and in this statement is nothing but outright vilification.' Goodman would translate the last two words as '...pure absurdity,' so that al-Ghazâlî would be calling the theologians' position absurd. Marmura, defending a more occasionalist reading of al-Ghazâlî, translates the same as 'sheer vilification,' referring to the philosophers. Marmura's is clearly the better translation (tashnî' being the verbal noun of 'to vilify'), indicating that al-Ghazâlî is in fact referring back to the criticism made by the philosophers. The passage is at Tahâfut, p. 296. See also Riker, p. 319.

(12) Tahâfut, p. 300, p. 258. English translation p. 330, p. 278.

(13) Tahâfut, p. 296. English translation, p. 325.

(14) Tahâfut, pp. 295-6. English translation, p. 324.

(15) Tahâfut, p. 294. English translation, p. 322.

(16) Qurân 35.43, cited at Tahâfut, p. 292. English translation, p. 320. See also Tahâfut, p. 302, English translation, p. 333.

(17) Tahâfut, p. 296. English translation, p. 325.

(18) Tahâfut, p. 298. English translation, p. 327.

(19) Tahâfut, p. 84. English translation, p. 70.

bluered.gif (1041 bytes)


Back to the Top

20th World Congress of Philosophy Logo

Paideia logo design by Janet L. Olson.
All Rights Reserved


Back to the WCP Homepage