Navigation – Plan du site

AccueilNuméros22Argumentation and Arabic Philosop...Analogical Arguments in the Kalām...

Argumentation and Arabic Philosophy of Language

Analogical Arguments in the Kalām Tradition: Abū l-Maālī al-Juwaynī and Beyond

Les arguments analogiques dans la tradition du kalām : Abū l-Maꜥālī al-Juwaynī et au-delà
Abdurrahman Ali Mihirig

Résumés

Cet article examine le développement et la critique des arguments analogiques dans la tradition du kalām. Il existe deux positions de base sur les arguments analogiques : (i) l'une soutient que si les arguments analogiques produisent une certitude, alors ils sont analysables comme des inférences déductives, rendant l'analogie elle-même redondante. En retour, ces figures soutiennent donc que si l'analogie est utile, elle ne donnera jamais la certitude exigée dans les sciences rationnelles ; (ii) une autre position soutient que l'analogie reste utile même lorsque l'argument est déductif, soit parce que la prémisse universelle est épistémiquement dépendante du cas source, soit parce que le cas source est utile dans des contextes dialectiques. L'exploration de ces thèmes sera centrée sur la formalisation par Juwaynī et la critique subséquente des arguments analogiques dans la tradition du kalām.

Haut de page

Texte intégral

1A commonly identified feature distinguishing ‘classical kalām from ‘post-classical kalām is the use of analogical arguments, known in the terms of the classical mutakallimūn as ‘analogy from the present to the absent’ (qiyās al-ghāʾib ala al-shāhid). Once Avicennian logic was integrated into the kalām tradition, as the story goes, they learned that analogical arguments did not yield certainty, and had to revamp their methodological apparatus. This article will follow some of these developments within the kalām tradition and the associated logical tradition on the question of analogical reasoning, taking Abd al-Qāhir al-Baghdādī’s (d.429/1037) classical work on argumentation theory (ilm al-jadal) as the starting point. In his Burhān, Abū l-Maālī al-Juwaynī (d.478/1085) subjects Baghdādī’s work to a strong criticism, especially on the question of analogical arguments and their utility in the rational sciences. I will first look at Juwaynī’s development of analogical arguments; second, I will examine his critique. Following this, I will examine its reception by his direct students Anṣārī and Ghazālī, in addition to later commentators: Muḥammad b. Alī al-Māzarī, Taqī al-Dīn al-Muqtaraḥ, Alī b. Ismāīl al-Abyārī, and Fakhr al-Dīn al-Rāzī. While there are grounds to accept the narrative that Juwaynī’s critique and the subsequent integration of logic displaced analogical reasoning in kalām, what we will actually find is that the classical modes of analogical argumentation, as formulated by Juwaynī, continued to be used even after he abandoned them himself. Beyond his particular formulation, figures such as Ṣadr al-Sharīa (d.747/1346-7) and Taftazānī (d.792/1390) insisted that analogical arguments could and did yield the certainty demanded by the rational sciences.

al-Baghdādī on Methodology in ꜥilm al-kalām

2According to Baghdādī, every science must have an aim (gharaḍ) and an instrument (āla). The aim of kalām is the following:

  • 1 The Arabic expression in the recently edited text is mā yajibu itimāduhu alā al-qaṭ wa-l-yaqīn, (...)
  • 2 al-Baghdādī, Abd al-Qāhir (2019), Iyār al-naẓar fī ilm al-jadal, Kuwait, al-Asfār, Aḥmad Arrūbī (...)

The desired aim from ꜥilm al-kalām and its rational foundations is only: arriving at knowledge of what is obligatory to believe with proof and certainty,1 without the permissibility of imitation in its roots or branches.2

The normative demands of this science that concern us here are the epistemic ones: proof and certainty (al-qaṭ wa-l-yaqīn), void of any imitation (taqlīd). Baghdādī is not unique in setting such a standard, but Juwaynī will judge that the methods he proposes to meet those aims simply fail to do so. It is for this reason that Juwaynī directs his critiques at Baghdādī, albeit without mentioning him by name. As for the instruments of kalām, Baghdādī states the following:

  • 3Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 151.

The primary instruments in ꜥilm al-kalām are two things: (i) proof (burhān), and (ii) refutation (ilzām). Proof is of two types: noninferential (ḍarūrī), and inferential (istidlālī). The noninferential is of two types: sense perception (ḥiss) and primary apprehensions (badīha). And inference is valid only by bringing [a proposition] back to sense perception and primary apprehensions. As for the methods of refutation (ilzām), they are many, and we shall mention them in the chapters on questions and objections.3

At the outset, then, kalām has two modes of argumentation. The first is known as burhān, or proof. The purpose of proof is to demonstrate one’s positive theses, while refutation (ilzām, lit., ‘causing or forcing an unwanted consequence’), is meant to demonstrate the falsity or absurdity of the theses of one’s opponents. Of course, in the event that logically there are only two possible positions, then the refutation of one will likewise be proof for the other. What matters to us here is that proof ultimately involves a type of epistemically tracing a proposition back to sense perception or first principles.

  • 4Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 155.

3The tracing back occurs by means of reflection (naẓar), which Baghdādī defines as the following: ‘Thinking (fikr) about the object of reflection with the aim of knowing its reality or one of its attributes, and [reflection] may lead to what is correct if it is ordered appropriately, and it may be erroneous, if it is not.’4 So far, the description is sufficiently vague. In a following chapter, Baghdādī presents us with additional details on what the structure of reflection is supposed to contain:

  • 5Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 180.

Everything which is known by the senses or intuitions, reflection (naẓar) has no role in validating it. Rather, what is known by reflection is only that which is absent from the senses and primary apprehension, if it contains something which entails that it be included with what is indicated by sense and primary apprehension.5

In other words, the general structure of reflection is meant to be the process by which one traces some inferential proposition which is absent to the self, that is, not known by sense perception nor primary apprehension, back to a proposition which is known by either one of them. The language of absence in his definition indicates further that the function of reflection is to acquire knowledge of the epistemically absent, i.e., unknown, in terms of the epistemically present, i.e., what is known.

4Often times, this is referred to as analogizing (qiyās) from the present to the absent. As such, it has often been understood as a straightforward analogy between what is present to what is absent, but this is a crude interpretation of the term that does not succeed in explaining many of the inferences that these authors actually held to fall under this category. Nevertheless, as we shall see in Baghdādī’s more detailed chapter on the specific inferences that are made in reflection, it involves more than the form described as ‘analogy between present and absent.’ It is this particular form of this inference, however, which does get formalized and then isolated for criticism by al-Juwaynī and others.

Inferential Forms of Proof

  • 6Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 282-284.

5Baghdādī divides his discussions of ‘proof’ into immediate proofs and mediated proofs.6 Baghdādī first describes three forms of immediate proof, which are interesting in their own right and likewise criticized by Juwaynī; nevertheless, it is in the following section discussing inferential proofs where we will focus our attention. Baghdādī here describes four forms of inferential proof. It is exactly these four, in the very same order, that al-Juwaynī will pick up on and criticize in his Burhān. Baghdādī writes:

  • 7Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 284.

As for proof that obtains through reflection and inference, it consists in four types: (i) one of them is inference from the known to the unknown; (ii) the second is building conclusions and secondary propositions on premises and primary propositions; (iii) the third is inference from agreement and difference; (iv) and the fourth is inference by division (or disjunction).7

Each one of these is then detailed by Baghdādī with examples. Although there are many interesting aspects in all of these inference forms, and they are all rejected or qualified by Juwaynī in the Burhān, we will focus on the inferences from present to absent.

Inferences from the Known/Present to the Unknown/Absent

6The first type of proof that Baghdādī presents is ‘inference from the present/known to the absent/unknown’ (al-istdlāl bi-l-shāhid alā al-ghāʾib). According to him, they are of two types:

  • 8Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 284.

One of them is the evidence (dalāla) of the present for something identical to it, such as the present fire’s evidence for the fact that the fire that is absent is identical to it in heat, dryness, and brightness; and this kind [of inference] falls under the evidence of the particular for the general (dalālat al-juzʾ ꜥalā al-kull).8

Instead of presenting us with a working definition of this type of inference, Baghdādī proceeds by division and example. The first type is an inference from known facts about some entity, to the fact that all identical beings will have the same relevant properties, only if they are properties which hold for that entity insofar as it is the same kind of thing. The example he gives us is fire. Thus, we can infer from the fact that all observed fires are hot, dry, and bright, that all unobserved cases of fire will be the same. This is the structure of the inference, but it remains to be seen what is the best way to analyze it; that is, is it deductive, inductive, or analogical?

  • 9  Bartha, Paul (2019), "Analogy and Analogical Reasoning", The Stanford Encyclopedia of Philosophy ( (...)

7In standard characterizations of analogical reasoning, an analogical argument is made by drawing similarities between the known object or system, referred to as the source domain, and the unknown object or system, referred to as the target domain. One states, for example, that S has the properties a, b, c, and d; and T has properties a, b, and c. Given that the properties a, b, and c, have some relevance for having the further property d, one infers that T (probably) has d as well. Arguments of this form are ampliative; since the conclusion is broader than what is strictly implied by the premises; depending on the nature of S and T, it is possible for the conclusion to be false while the premises remain true.9

8But it is not clear that this is the appropriate way of analyzing the argument presented by Baghdādī in his example. The structure of this argument appears to be something like the following: (i) By observation we come to know that fire is hot, dry, and bright; (ii) therefore, we conclude that unobserved fires are hot, dry, and bright. More abstractly, we could think of this inference as the following:

If the properties x, y, and z, are caused by, or concomitant to, the reality q, in all observed cases (the source domain), then whenever q exists, it shall cause or be concomitant to the same properties x, y, and z in all unobserved cases (the target domain).

We can interpret this in different ways. One way is to suggest that this is not an analogy at all, but rather, a generalization which is then applicable or transferred to unobserved cases. It is not even clear that we need to have unobserved cases if we interpret the statement as a conditional. In other words, it is a kind of inference that can make predictions about unobserved cases in both time and place; no direct comparison is needed because the inference follows from the very reality of the thing being observed. On this reading, the analogy turns out to be a straightforward deduction.

  • 10 This is not necessarily the case of course. It may be instead that one identifies an essential prop (...)

9On the other hand, we could also understand Baghdādī as presenting an inductive argument. That is, it could be read as a generalization from a single or limited set of instances, to all instances, and therefore, to any particular instance.10 In the case that our conclusion is all cases, observed and unobserved, it would be an inductive inference; in the case that our conclusion is a particular case or set of cases, then it would be analogical.

10Textual support for this inductive interpretation can be found in the statement he makes at the end of his example: ‘this kind [of inference] falls under the evidence of the particular (juzʾ) for the general (al-kull).’ While Baghdādī uses the language of ‘part’ and ‘whole’, what he actually means by juzʾ is a particular, and what he means by kull is ‘all.’ If this is right, then it could plausibly be read as an inductive argument, since one is arguing from the truth of an individual or particular proposition, to a general, universal proposition. Likewise, one could understand inductive arguments as a form of analogical argument; that is, by analyzing some limited set of observed samples, one then analogizes or extends the judgment to all other unobserved objects. Nevertheless, Baghdādī’s language is sufficiently vague that one could also interpret this type of inference as being a deduction, namely, by identifying an essential property in one individual, and from there, judging that all individuals have the same property.

11Interestingly enough, Baghdādī’s predecessors al-Asharī and al-Bāqillānī reject this form of inference using the exact same example of fire:

  • 11 Ibn Fūrak, A. B. M. b. al-Ḥasan (1987), Mujarrad maqālāt Abī l-ḥāsan al-Asharī, Beirut, Dār al-Mas (...)

If someone says: Is it not the case that since you have not observed a fire that is not hot, nor an agent that is not a body, that you then judge that every agent is a body, and that every fire is hot? He is answered by stating: We do not come to know fires absent to us by inference from present fires. Rather, we know that whatever the name of fire is true of, be it present or absent, is something which is hot, by virtue of linguistic convention, in that everything that is of this heat, light, and structure, is fire, because that is their convention…So our position on present fire is the same as absent fire, and we do not say that absent fire is hot because present fire is hot.11

  • 12 Which for Asharī is at root a Divine Convention, cf. Mujarrad, p. 41-42.

In other words, Asharī is stating that we have a concept of fire that we acquire through the acquisition of language.12 It is neither our choice nor our inference what the essential properties of fire are: we have inherited a practice of asserting that whatever is bright, hot, and of a certain structure, is called fire. Asharī appears to be resistant to the idea that mere observations of certain properties are enough to extend that judgement to all unobserved cases. Baghdādī’s language above appears to be interpretable in both ways.

  • 13Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 284.

12Baghdādī’s second type of ‘inference from known/present to unknown/absent’ is “the evidence of the present for what is different from it; such as the evidence of an action for the existence of an agent, and the evidence of writing for the existence of a writer.”13 In this case, the inferential structure of the argument appears to have a straightforward deduction that is premised on some general or limited form of the causal principle. Let us analyze the two examples:

  1. There exists an act (fil)

  2. Every act has an agent

  3. Therefore, this act has an agent.

13The second example is just a more limited case of the first:

  1. A piece of writing exists

  2. Every piece of writing has a writer

  3. Therefore, this piece of writing has a writer.

If this analysis is correct, then the second form of ‘inference from the known to the unknown,’ is a straightforward deductive inference. In both cases we infer from some entity or event, such as a creative act (i.e., the existence of any contingent thing) to the existence of something that is different from it, namely, an agent. Likewise, a piece of writing is a different kind of being from writers.

14Since Baghdādī appears to consider both divisions as sub-types of a single kind of inference, it is probably best to consider them both as deductive inferences. In the first class of inferences: we infer from a single individual to the whole kind. We observe one fire as being hot essentially, then we judge that all individuals are hot as well. The reason why Baghdādī speaks of ‘particular’ and ‘whole’ is likely because he does not seem to admit universals into his ontology. As such, what appears to be an inductive inference because he is speaking about inferring from one particular to all particulars, is in fact, a deductive inference from one particular to the universal (i.e., all particulars) through the medium of an essential property, as suggested in the initial formulation above.

15In the second class of inferences, the inference from an entity to what is different from it, we have another deduction. The logic appears to be something like the following: every entity we observe implies some set of necessary consequences, concomitants, or entailments. Some of these concomitants or entailments pertain to the thing itself, namely, when we identify properties which are necessary of the thing. Then there are properties which the entity has which also entail by necessity the existence of something different and external to it. For example, contingent things by necessity imply a cause which is external to them. A piece of writing, likewise, possesses properties which by necessity imply the existence of a writer that is external to it. Therefore, it is very plausible to read both inferences as deductive inferences, and, furthermore, to understand that these deductive inferences are either an inference to an internal or intrinsic fact, or they are an inference to an external or extrinsic fact.

16It is interesting to note that the main difference that Baghdādī highlights between the two inferential forms is ontological in nature; that is, in one inference, we come to knowledge of the very thing we are investigating, and in the other inference, we come to knowledge of something distinct from the object of our investigation. As we shall see, this appears to be motivated at least in part by disputes with other sects and philosophical schools which claimed, for example, that we cannot come to have knowledge of metaphysical questions. But if it is possible to infer from the existence of something to the existence of something different from it, then this opens the possibility of inferring from the temporal to the eternal.

17Lastly, Baghdādī illustrates the latter point in a later section on the disagreement over this form of inference, from the known to the unknown. He writes the following:

  • 14Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 297.

They differed over the inference from the present to the absent and its procedure, in three opinions: (i) the Manicheans held that the present can only indicate its like in the absent, and they called this ‘the evidence of the part/particular for the whole/universal’; (2) the eternalists among the ‘atheists’ held that the present may indicate what is beyond its whole, but it can only indicate its like; such as a house that we observe being built by a builder, then we know he is its builder. Then we observe another built house, then we know that its builder must be similar to the builder of the first house; (3) the Party of Truth said: the present indicates its like as we mentioned before, and it may indicate what is different from it, such as writing indicating the existence of a writer, and smoke indicating fire, and so on.14

  • 15Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 297-8.

Baghdādī then presents two dialectical arguments against the two opponents. Against the Manicheans he states that if they only accept that like signifies like, then they should have no grounds to infer from the mixed bodies of the world to the existence of a body of pure light and pure darkness.15 That is, since they claim that we can only make inferences of the first type discussed above, they no longer have any epistemic justification for the existence of the two substances of pure light and pure darkness, which they claim are the two constitutive elements of reality. The only kind of inferences they should be justified in making are ones about mixed bodies only, since we observe neither pure light nor pure darkness in the world.

18Meanwhile, the Eternalists argued that since every human being we have observed has proceeded from a prior human being (and other animals from like animals, and so on), then, we are justified in generalizing this statement to the following: every originated entity derives or originates from a previous entity, and conversely, creation ex nihilo is impossible.

  • 16Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 298.

19Baghdādī responds by stating that we observe that every accident that occurs in observable reality occurs without an origin, and on the reasoning of the eternalists, we should infer that, therefore, every event that occurs comes into existence ex nihilo, i.e., from no origin (aṣl).16 But since such an extension of the judgement is rejected by both parties, there must be something arbitrary in the inference itself. Presumably then, both of these examples show that there are valid and invalid ways to infer from the present to the absent. Baghdādī, however, does not explicitly point out where his opponents go wrong, but only that something in the inference has gone wrong. As we shall observe later very clearly in Juwaynī, arbitrary inferences are presented as the main motive behind formalizing inferences from the known/present to the unknown/absent.

Abū l-Maꜥālī al-Juwaynī in the Burhān

  • 17 Tāj al-Dīn al-Subkī (n.d.) Ṭabaqāt al-Shāfiiyya al-kubrā, Cairo, Dār Iyāʾ al-Turāth al-Arabī, Mu (...)

20Juwaynī’s Burhān was famously described by the Shafiī jurist and historian Tāj al-Dīn al-Subkī as ‘the riddle of the community’ (lughz al-umma).17 One reason for this is that Juwaynī spends some time covering a number of difficult topics in a very critical yet obscure manner. For example, he rejects earlier definitions of intellect (aql), knowledge (ilm), and reflection (naẓar); he presents a controversial view on God’s knowledge of an infinite number of objects; he rejects the theory of abstract modes (aḥwāl), which he had previously defended; and most relevant to our purposes here, he rejects the 4 classical modes of proof maintained by his predecessors, such as Baghdādī.

21Juwaynī presents us with a programmatic statement first:

  • 18 Abū l-Maālī al-Juwaynī (1978), al-Burhān fī uṣūl al-fiqh, Qatar, Qatar University, Abd al-Aẓīm a (...)

Our predecessors ordered the proofs of reason in a particular order that we shall transmit here, then we shall demonstrate its falsehood and present our chosen alternative, and in so doing, we shall combine the transmission of previous views and indicate what is true in them.18

Immediately after this introductory statement, Juwaynī cites the four modes of argument that we saw in Baghdādī:

  • 19 Baghdādī
  • 20Abd al-Malik al-Juwaynī (1978), Burhān, 126; cf. Abd al-Qāhir al-Baghdādī, (2019), Iyār, p. 284.

They said: the proofs of reason divide into four: one of them is basing the absent on the present, the second is basing conclusions on their premises, the third is collection and division,19 and the fourth is inferring from the point of agreement to the point of disagreement.20

Aside from the fact that Baghdādī makes ‘collection and division’ the last of the four, and Juwaynī here makes it the third, there is no other difference. Nevertheless, in his actual discussion of these four modes, Juwaynī assumes the same ordering that Baghdādī does. He first elucidates the inference from the present to the absent, then from the premises to conclusions, then the inference from agreement to disagreement. He does not mention disjunction and division until the end of his critique of the previous three. Notably, Juwaynī’s formalization of the inference from present to absent is not the same as we find in Baghdādī. It rather seems to be Juwaynī’s own unique formalization which, as we shall see, becomes the standard formalization of the inference from present to absent.

Juwaynī on Inferences from the Present to the Absent

  • 21Abd al-Malik al-Juwaynī (1981), al-Shāmil fī uṣūl al-dīn, Tehran, Tehran University, R. M. Frank ( (...)

22Juwaynī’s description of this mode of inference is summarized from his longer kalām work, al-Shāmil fī uṣūl al-dīn; a succinct version of this inference is also found in his widely studied Irshād.21 Whenever appropriate, the reading of the Burhān will be supplemented with references to the Shāmil and the Irshād. We see in Juwaynī a formalization of this kind of inference that is not present in earlier authors. As we shall see in the reception of Juwaynī’s work, most authors credit him with being the originator of this analysis of the inference from present/known to absent/unknown.

  • 22Abd al-Malik al-Juwaynī (1978), Burhān, p. 127.
  • 23Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 350.

23The first point that Juwaynī raises is that “it is not permissible to make arbitrary analogies without a rationally discernible common element (jāmi aqlī).”22 The term “rational” here has connotations in kalām; rational items are those which are necessary, universal, and independently verifiable by rational agents without recourse to tradition, convention, or revelation. There are no ‘exceptions’ to a rational principle of any kind; if the property holds, then it holds by necessity. For example, if it is rationally the case that every temporally originated event needs a cause, then it is necessarily the case that it needs a cause.23

  • 24Abd al-Malik Juwaynī (1978), Burhān, p. 127; cf. Abd al-Malik Juwaynī (1981), al-Shāmil, p. 60-61 (...)

24In the case of making an inference from the known/present to the unknown/absent then, one must identify a common rational element that exists in both domains. Furthermore, that rational element must be the item which explains the further property we are trying to infer in the target domain. Arbitrary cases, according to Juwaynī, will lead us to false conclusions. In the Burhān) he mentions two examples: (i) the anthropomorphists (al-mushabbiha) argued that ‘We have never observed an agent that was not physically determined,’ and therefore, they infer that God must be a physically determinate entity. (ii) The atheists (al-muaṭṭila), on the other hand, claimed that ‘an existent which has no similarity with any existent we know of is unintelligible,’ and therefore, they denied the existence of God.24

  • 25Abd al-Malik al-Juwaynī (1981), al-Shāmil, p. 61.
  • 26Abd al-Malik al-Juwaynī (1981), al-Shāmil, p. 61.

25In his longer work al-Shāmil, Juwaynī cites further examples of these arbitrary generalizations: (i) the ‘atheists’ (al-mulḥida) stated: ‘we have not observed a plant except from a seed, nor a chicken except from an egg, nor a planetary orbit except that it was preceded by another, and therefore, it is necessary that we extend that judgment to all that we have not observed.’ Therefore, one would conclude the world is eternal.25 (ii) The Manicheans (al-thanawiyya) stated that ‘we have found that all bodies divide into the dark and the light, and therefore, we must judge accordingly on all that we have not observed.’ Therefore, there must be some distinct origin for each. (iii) The naturalists (al-ṭabāiiyyūn) stated that ‘we have found that all bodies are never void of the four elements’; (iv) the corporealists held that ‘We have not observed a self-subsistent entity except that it occupies some specific location in a direction.’ Juwaynī then comments ‘The corporealists have no basis for this except for the arbitrary reference to the observed.’26

26What we notice in all these examples is the same type of error: a generalization from observed cases without identifying the common basis behind the common judgment. In the Burhān, Juwaynī takes this as a given without much explanation. In the Shāmil, however, we get some further details of what is wrong with these inferences. Juwaynī first presents a series of refutations of each of the positions he has mentioned, albeit briefly, in showing that each one of them comprises an inconsistency. Beyond that, Juwaynī says the real source of their error is the following:

  • 27Abd al-Malik al-Juwaynī (1981), Shāmil, p. 62.

We say to them: the upshot of your statement amounts to saying that whatever we have not observed must be the same as what we have observed, and this is completely arbitrary; for it is not necessary that what is absent to the senses must be like the sensible, nor is it necessary to assume that the observations of a single observer encompass all describable entities.27

Juwaynī goes on to mention two scenarios which exemplify the charge of arbitrariness:

  • 28Abd al-Malik al-Juwaynī (1981), Shāmil, p. 62. It would appear that this thought experiment forms (...)

It would be necessary for one who was raised among black Africans to judge that all human beings had the same attributes as the black Africans; and it would be necessary for one who had never seen death, nor heard of it, and having been placed on island as an infant, then lived there alone, then it would be necessary for him to die in a state of being certain [that there is no death].28

From all these examples, it is plausible that the kind of analogical argument at play here may be analyzed as (i) a generalization from known cases; then (ii) applying that generalization to the target case by identifying the relevant common element. The first element of Juwaynī’s analysis, therefore, is how to prevent arbitrary generalizations or arbitrary applications of those generalizations to new cases. Specifically, the kind of inference that Juwaynī rejects is the following:

  1. All observed cases of x have been y.

  2. Therefore, all unobserved cases must also be y.

This kind of inference, according to Juwaynī, is arbitrary. The fact that all observations have been one way or another, is never evidence alone to determine that all unobserved cases must be that way as well. In the absence of a common element which guarantees the judgement, there is no basis for generalization and extension to new cases. In other words, Juwaynī wants the argument to be improved to account for the common element that ties the two domains together, what he refers to as the jāmi. We may describe it as follows:

  1. Every x is y because of c. (source domain)

  2. c holds for z (the target) (by virtue of an external argument).

  3. Therefore, z is also y, because of c (applying the generalized inference to the new case).

  • 29Abd al-Malik Juwaynī (1978), Burhān, p. 127-128; cf. Abd al-Malik Juwaynī (1981), Shāmil, p. 63-6 (...)

What grounds the inference, therefore, is the presence of the common element c. Which kinds of elements, however, are candidates for the validity of this inference? Juwaynī tells us that there are four: (i) cause (illa); (ii) reality (ḥaqīqa); (iii) condition (sharṭ); and (iv) evidence or signifier (dalīl).29 All but one of these factors are identified in Ibn Fūrak’s Mujarrad, where we find the following:

  • 30 A. B. M. b. al-Ḥasan Ibn Fūrak (1987), Mujarrad, p. 80.23-25.

[Ashꜥarī] used to say: The observable domain alone has no evidentiary value in matters of analogy, inference, and deriving testimony from the observable to the unobservable except after (one identifies) some added signification (dalāla), thus, distinguishing between descriptions, causes, conditions, and realities (awṣāf, ꜥilal, shurūṭ, ḥaqāʾiq).30

  • 31 The term dalāla above is not identified as common elements. Instead, each one of those 4 identified (...)
  • 32Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 129-131.

Aside from the fact that ‘dalīl’ is not mentioned here, it is not difficult to see how one can abstract from analyses of this time to the formulation found in Juwaynī.31 Nevertheless, it appears that no extant texts have taken the further step that we find in Juwaynī. Likewise, Baghdādī, who claims to be offering the first comprehensive work of Jadal, does not offer a similar description of inference from present to absent.32 As we saw earlier, he divided them into two types only: inferences from a present particular to absent particulars of the same kind, and inferences from a present particular to an absent particular of a different kind. Since Juwaynī’s purpose in the Burhān is to reject all these modes of inference, he is satisfied with a brief treatment by way of example only. As such, it is more useful to look at the more detailed treatment in the Shāmil before looking at his critique in the Burhān.

The Cause as Common Element

27The argument from common cause (illa) is the following:

    • 33Abd al-Malik al-Juwaynī (1981), Shāmil, p. 63.

    If a known judgment j holds of a subject s by virtue of cause c, then it must be the case that whenever that same judgment j is true of the target subject t, then t must also have the cause c.33

  • 34Abd al-Malik al-Juwaynī (1978), Burhān, p. 128.

In the Burhān, Juwaynī gives us an example of this inference: ‘If being a knower (kawnuhu ālim) is caused by knowledge (ilm), then it is necessary to extend that to the unknown/absent.’34 This was important for arguing for the existence of real attributes in God: (i) one first argues that the creator must be knowing, by virtue of the observation of precision and design in creation; (ii) then one argues that the property of ‘being a knower’ is true if and only if the agent has knowledge, then it follows that the Creator has the attribute of knowledge. It is in this second part of the argument that we extend the judgement. Here one may ask: why is it the case that ‘being a knower’ is true if and only if the agent has knowledge? If the answer is merely, ‘Because all the cases of knowers we have observed have knowledge,’ then the inference is guilty of arbitrary extending the observable to the unobservable without reference to a common element.

  • 35Abd al-Malik al-Juwaynī (1981), Shāmil, p. 63.

28Juwaynī entertains precisely this objection in the Shāmil.35 In response, he makes the following argument:

  • 36 This is a previous book from within the Shāmil, which has been partially edited in the edition by N (...)
  • 37 By the term biconditional, I mean the necessary co-presence and co-absence of the two elements.
  • 38Abd al-Malik al-Juwaynī (1981), Shāmil, p. 63. The term ḥukm is often translated as ‘judgement’ or (...)

This is not the same case as we argued before, for we have clarified in the Book of Causes (kitāb al-ꜥilal)36 that if a property (al-ḥukm) obtains by virtue of a cause in what is known, or, that proof has been established that there is a necessitating cause for [the judgment], then that cause is not realized except in a strictly biconditional37 way (illā maꜥa al-iṭṭirād wa l-inꜥikās); nor is the effect of the cause established except with the certainty that the property (ḥukm) cannot obtain without it, just as [the cause] does not obtain without being effective…thus, negating causal dependence in the property in what is unknown (the target domain) implies negating it from the property in what is known (the source domain); but its [causal dependence] has been established in what is known, and by this, we evade the objection.38

29What Juwaynī argues here is that the dependent property (ḥukm) of ‘being a knower’ only obtains because the subject possesses the grounding property (illa) of knowledge. Whenever one property necessitates or grounds another for its subject, it is called a illa, or a cause. Thus, we can infer from the dependent property – which is itself established by external evidence – to the existence of the grounding property which entails it. If we do not apply this inference in the case of God, for example, then we are basically stating that ‘being a knower’ can hold without knowledge, and by extension, our inference in the source domain is also falsified. But since this is a metaphysical impossibility in the observable, it must also be impossible in the unobservable.

30Furthermore, as Juwaynī argues in Kitāb al-ilal, this inference is grounded in the theory of aḥwāl (abstract modes). At one point he writes:

  • 39Abd al-Malik al-Juwaynī (1969), Shāmil, p. 631.

If someone says: Clarify the truth for us between the two positions (i.e., affirming or negating the abstract modes), and abandon this hesitation in the discussion. We say: What is stronger according to us is the affirmation of the modes, and this is a question of immense importance, upon which all the properties of causes (ꜥilal) depend, and they are the foundations of evidence, and the sources of definitions.39

  • 40 For more on the aḥwāl, see Fedor Benevich (2016), “The Classical Asharī Theory of Aḥwāl: Juwaynī a (...)

An adequate discussion of the theory of aḥwāl is beyond the scope of this paper, but it is important to give it a brief treatment.40 The theory of abstract modes is one which is functionally similar to Aristotle’s theory of universals. They are dependent, secondary properties of real entities that account for the commonalities between things. For Juwaynī, they are the basis of all arguments and definitions too, since, without some extramentally real (albeit ‘not independently existent’) properties that are shared between entities, then a single definition cannot pick out all the intended items subsumed by it, and likewise, predicates cannot be transferred between premises in arguments. Suppose Zayd has knowledge of theology, while Amr has knowledge of grammar. It is plain to see that Zayd’s individual knowledge of theology is distinct from Amr’s individual knowledge of grammar; they are numerically distinct and qualitatively distinct. Nevertheless, the existence of these two individual properties in Zayd and Amr respectively makes it true to say: ‘Zayd is a knower’ and ‘Amr is a knower’. These shared, i.e., universal, predicates, according to abstract-mode theorists, are the aḥwāl. They represent, as it were, the ‘indigenous’ kalām theory of universals.

  • 41Abd al-Malik al-Juwaynī (1969) Shāmil, p. 646.
  • 42Abd al-Malik al-Juwaynī (1969), Shāmil, p. 644-645. Of course, this inference could be denied by t (...)

31For Juwaynī, an abstract mode is grounded either in an essence (a dhāt) or a real property (a ma). The ones we are concerned with are the ones instantiated by a property. Indeed, he defines a cause (illa) as the following: ‘a real attribute which necessitates a further property (ḥukm) for the [subject] in which it subsists.’41 Juwaynī will go on to argue that our knowledge of the dependent property (the ḥukm) is distinct from our knowledge of its entailing cause (the ma or illa), and if the knowledge is distinct, then its object must be distinct. But if the judgment is merely a different way of expressing the same fact – the fact that the subject has the grounding property – then this would be insufficient grounds to explain the distinctness of the knowledge of each. For example, Juwaynī would hold that our knowledge of: (i) Zayd’s knowledge of grammar, is distinct from our knowledge of (ii) Zayd’s being a knower. Thus, the only way to make sense of the fact that our knowledge of each one is distinct is by postulating a distinction in external reality, namely, by distinguishing between the grounding property of knowledge (the illa) from the entailed abstract mode or general/universal property (i.e., the ḥāl) ‘being a knower’. In turn, it is only by means of this metaphysical scheme – according to Juwaynī – that we can make sense of inferences from the truth of certain judgments to the existence of certain properties.42

32Having explained the metaphysical structure that grounds the argument by common cause, we can now attempt another characterization of the inference:

  1. If the abstract property j is necessitated by the grounding property c in the source domain, then it is necessarily the case that an analogous abstract property jt is necessitated by a similar grounding property ct in the target domain.

Therefore, if we establish the fact of the abstract property j in the target domain, and we know that such properties obtain if and only if it is necessitated by the grounding property c in the source domain, then it must be the case that the abstract property jt in the target domain is also necessitated by a similar grounding property c. To be sure, such an inference would be necessary, since by the agreement of all aḥwāl theorists, the abstract modes have no existence or reality on their own. They are either grounded in an essence, or they are grounded in an additional property. In both cases the ‘existence’ of the abstract mode just is the existence of the essence or grounding property respectively. Therefore, if the relevant grounding entity is absent, so is the abstract property. Then, since both parties concede that ‘being a knower’ is a mode grounded in the property of knowledge in the source domain, they must concede a grounding property of knowledge in the target domain if they affirm the abstract property of ‘being a knower’.

Reality as Common Element

  • 43Abd al-Malik al-Juwaynī (1981), Shāmil, p. 64.

33The second relevant common element is ‘reality’ (ḥaqīqa). The reality of something, according to classical kalām, is roughly what is meant by the term essence. The general form of the inference, given by Juwaynī in the Shāmil, is the following: ‘The reality of x in the observable/known domain is y, therefore, the reality of x is y in what is unknown/unobservable.’43 Therefore, if the reality of a knower, i.e., its definition, is ‘that which possesses knowledge,’ then one can make the following argument:

  1. God is a knower (by some external argument)

  2. Every knower is a being which possesses knowledge (by virtue of the reality of ‘being a knower’).

  3. Therefore, God possesses knowledge.

  • 44Abd al-Malik al-Juwaynī (1978), Burhān, p. 128; Abd al-Malik al-Juwaynī (1981), Shāmil, p. 64.

The ‘analogical’ element in the argument is implied in premise two: if it is the case that ‘being a knower’ in the present/observable domain is one who possesses knowledge, then it must be the case that all unobserved cases of knowers must also possess knowledge. In order to avoid the charge of an arbitrary generalization, this argument relies on the presence of a common reality (ḥaqīqa). That is, from examining the notion of ‘being a knower’ in the observable domain, one deduces that it means to possess knowledge. Then, if some evidence indicates that a being in the unobservable target domain is also a knower, then it follows that this being must also have knowledge. In the case above, it is the reality of a ‘knower’.44

Condition as Common Element

  • 45Abd al-Malik al-Juwaynī (1981), Shāmil, p. 63; cf. Burhān, p. 128.
  • 46Abd al-Malik al-Juwaynī (1978), Burhān, p. 128.

34The third candidate for a common element is ‘condition’ (sharṭ). The general form of this inference is given by Juwaynī as follows: ‘If a judgment is conditioned by some condition x in observable reality, then it is conditioned by that condition x in unobserved reality.’45 In the Burhān, Juwaynī is satisfied with an example: ‘It is like our statement: knowledge is conditioned by life in observable reality, therefore, it is necessary to judge the same for unobservable reality.’46 We can formulate this argument as follows in the case of God:

  1. Life is a necessary condition for knowledge in all observable cases.

  2. If life is a necessary condition for knowledge in all observable cases, then life is a necessary condition for knowledge in all unobservable cases.

  3. God has knowledge.

  4. Therefore, God has life.

Just as we saw in the previous two cases, in the argument for the second premise, the theologian needs to demonstrate the necessity of this condition for knowledge; and if it is a necessary condition, then it follows that all instances of knowledge must also be conditioned by it. Thus, the generalization is inferred from the present/observable domain, then it is extended to the target domain by virtue of the common element, namely, the condition. The main difference between the condition case and the reality/cause cases is the following: (i) one can infer from the conditioned property, such as knowledge or perception, to the existence of the condition, such as life but not vice versa, i.e., inferring from life to knowledge, because the condition is necessary but not sufficient for the existence of the conditioned; while (ii) one can infer from the causal property/reality to the caused property, and vice versa. Thus ‘knowledge’ is a necessary and sufficient condition for ‘being a knower.’

‘Evidence’ as Common Element

  • 47Abd al-Malik al-Juwaynī (1981), Shāmil, p. 64.
  • 48Abd al-Malik al-Juwaynī (1981), Shāmil, p. 64. That is, unlike the first two cases which are bicon (...)
  • 49Abd al-Malik al-Juwaynī (1978), Burhān, p. 128.

35The final candidate for a common element is ‘evidence’ (dalīl). Juwaynī’s general formulation of this inference in the Shāmil is the following: ‘Everything which signifies x in observable reality, must also signify x in unobservable reality.’47 This is necessarily the case, he argues, because the condition for something to have rational evidentiary significance is that it does so by itself, that is, essentially; nevertheless, Juwaynī admits that, unlike the previous three cases we discussed, the absence of evidence does not signify the nonexistence of the signified (madlūl).48 Therefore, since the significance of a given piece of evidence is essential to it, then it must signify the signified universally, or not at all. Thus, since it does in fact signify the signified in the source domain, we must also assert its evidentiary value for the target domain. Juwaynī gives us an example in the Burhān: ‘As in our statement: ‘temporal origination, specification, and precision, signify power, volition, and knowledge in observable reality; therefore, it is necessary to extend that judgment to unobservable reality.’49 In other words, in the observable source domain: the temporal origination of an entity signifies by rational necessity the existence of a cause with power (qudra); the specification of one set of possibilities over other equal possibilities indicates the existence of volition (irāda); and precision in design indicates the existence of knowledge (ilm). Therefore, the same events must signify the same properties in the target domain.

Juwaynī’s Critique of the Analogical Form in the Burhān

36Before coming to Juwaynī’s critique in the Burhān, it is useful to examine his general anti-skeptical argument for all forms of inferences from the known/present to the unknown/absent. In the Shāmil he states the following:

    • 50Abd al-Malik al-Juwaynī (1981), Shāmil, p. 65.

    Every judgement whose existence is tied to something in the known/present domain is such that if it were not tied to it in the unknown domain, it would imply the vitiation of what is known in the known/present domain.50

That is, if we are inconsistent in our application of certain rational inferences, namely, by applying them in the source domain and denying them in the target domain, then we end up undermining all of our inferences in the source domain as well. In other words, if we have come to have knowledge of a certain proposition in the observable-source domain, because of some necessity that holds between the subject and the predicate, then it is necessary to extend that judgment to the target domain by virtue of that relation. If we do not do so, then we are in fact conceding that the relation in the observable-source domain is not necessary, undermining all of our inferences there as well. Let us now turn to Juwaynī’s critique in the Burhān.

  • 51Abd al-Malik al-Juwaynī (1978), Burhān, p. 129-131.

37Juwaynī begins with a critique of the analogical form of reasoning; then he critiques Baghdādī’s noninferential, inferential, and ‘logical consequence’ inferences; then he critiques the inference from agreement to disagreement; then he begins his discussion of division and disjunction.51

38As for analogical arguments, Juwaynī states first that (i) all are agreed that naïve or arbitrary generalizations are to be rejected at the outset. As for arguments with a relevant common element, Juwaynī writes the following:

  • 52Abd al-Malik al-Juwaynī (1978), Burhān, p. 129-130.

As for commonality by means of cause (ꜥilla), it has no basis; for according to us, there is no cause nor effect, for a knower, being a knower just is knowledge. And commonality by reality is nothing, for temporal knowledge is entirely distinct from eternal knowledge, so how can they participate in a single reality despite this (essential) difference? If it is said: they both share in ‘knowledge-ness’ (al-ꜥilmiyya), then this is false, for it is based on the theory of aḥwāl, and we shall clarify its falsehood as needed.52

In this brief excerpt, Juwaynī rejects two of the four common elements discussed above: common cause (illa) and common reality (ḥaqīqa). One may have expected that his rejection of the inference from common cause would be the result of his rejection of the theory of abstract modes, but at least here he appears to say something different. That is, he is rejecting the very idea that there is an ontological dependence of the grounded property (malūl) on the grounding property (illa). But it would appear that he had already qualified this relation in the Shāmil by suggesting that in the created domain, it is not that the grounding property causes the grounded property in the productive sense, but that it implies it by necessity, but God remains the cause of both simultaneously:

  • 53Abd al-Malik al-Juwaynī (1981), Shāmil, p. 28.

But what [al-Bāqillānī] settled on in the Hidāya and the work entitled ‘The Caused and the Uncaused’ (mā yuꜥallal wa mā lā yuꜥallal) is that these properties, if we affirm them as abstract modes (aḥwāl), then all of them obtain by an agent (bi-l-fāꜥil), and the grounded property of a grounding property also obtains by an agent. For our intention in stating ‘the grounding property (al-ꜥilla) necessitates the grounded property (al-maꜥlūl)’ is not that it produces it just as power causes the occurrence of its objects; but rather we mean by ‘necessitate’ the implication (talāzum) of the grounded property and the grounding property, and the impossibility of either one obtaining without the other.53

It is therefore clear that the previous inference that Juwaynī asserted in his Shāmil and his Irshād was not dependent on the actual productivity of the ‘cause’ on its ‘effect’; all that mattered was the impossibility that one obtains without the other, such that we can always infer from one to the other and vice versa. Therefore, there appears to be a problem with his statement above. Nevertheless, the rejection of the abstract modes would eliminate this inference all the same, and so his comment on the denial of cause and effect is ultimately of no consequence.

39As for common cause, the theory outlined above assumed a difference between the entailed property j (the ḥukm), and the cause of that judgement, c. This distinction, according to Juwaynī in the Shāmil, is in turn dependent on the affirmation of the abstract modes. In other words, the judgement refers to the abstract mode which is a real property that is distinct from the real property c that grounds it. It nevertheless ‘does not exist’ in the sense that it has no existence of its own; rather, its existence is identical to the existence of its grounding cause. Looking at the inference again:

  1. If the grounded property j is true of subject S, then it is because S has the grounding property c.

  2. The grounded property j is true of the subject S.

  3. Therefore, S has the grounding property c.

But once Juwaynī has abandoned his view of the theory of modes, there is no longer any metaphysical difference between the grounded property of, e.g., ‘being a knower,’ and the existence of the grounding property ‘knowledge’. That means we can replace the statement ‘the judgement j is true of S’ with ‘the property c holds for S’ Therefore, the inference would look like this:

  1. The property c holds of S if and only if S has the property c.

But this is now trivially true and can no longer be used to infer one thing from another. In other words, collapsing the distinction between the grounded property and the grounding property by rejecting the metaphysical difference between them renders the inference from judgment to cause impotent.

40As for the case of the ‘reality’ or essence as common element, Juwaynī rejects that this can be of any use when it comes to theological questions (which means that he is not averse to using this inference within the temporal source domain, so it is not a complete rejection of this kind of inference). In keeping with the affirmation of divine transcendence, there can be no participation of a real property between Creator and creation. Therefore, to argue by a common reality – i.e., a universal property such as an abstract mode – does not work. However, this does not eliminate the efficacy of the same inference for like beings. Thus, Baghdādī’s argument from the observation of one or more fires as being essentially hot, to the fact that all fires are hot, is not something that would be vitiated by Juwaynī’s criticism here.

41Nevertheless, Juwaynī has a more general criticism of the inference from present domain to absent domain, which is the following:

  • 54Abd al-Malik al-Juwaynī (1978), Burhān, p. 130.

The general statement on this matter is that if evidence has been established for a desired conclusion regarding the absent/unknown, then that is the aim; and mention of the known/present has no relevance whatever. And if evidence has not been established for a conclusion regarding the absent/unknown, then mentioning the present/known case has no meaning either. Furthermore, there is no analogy in rational matters; and this applies to condition and evidence (i.e., as common element).54

Juwaynī here is making the following argument: either one’s evidence indicates the desired conclusion in the given domain, or it does not. If the evidence indicates the conclusion, then any discussion of the source domain, i.e., the present or the known case, is redundant. That is, if there is sufficient evidence to prove one’s case, for example, that God has knowledge because He is a knower, then mentioning the fact human beings have knowledge because they are knowers is irrelevant. Conversely, if we have no evidence that God has knowledge, then the human case likewise is of no use at all.

42To summarize, Juwaynī presents two specific arguments against the analogical inference by common cause and reality; and one general argument against all forms of ‘analogical’ arguments. (1) Juwaynī rejects the inference from cause (illa) because he denies any ontological status to the abstract modes which were the effects of those causes. This renders the inference impossible because it is redundant. Thus, one cannot infer from ‘being a knower’ to the existence of knowledge, because ‘being a knower’ just is ‘knowledge’. In other words, to demonstrate them is not to use the inference from present to absent, but rather, to infer from precise and orderly actions to the existence of knowledge. (2) Juwaynī also rejects the inference from ‘reality,’ at least in theology, because there are no common properties between the source domain (human beings) and the target domain (God).

43(3) The general logical critique of all such inferences is that mentioning the known/present is redundant. What matters, according to Juwaynī, is whether the evidence indicates the conclusion at all, irrespective of whether it is in the present/temporal or absent/eternal. Therefore, (i) if the evidence indicates a conclusion in the eternal domain, then that is the aim, and mention of the temporal adds nothing to the inference; and if the evidence does not indicate such a conclusion, then similarly, the fact that it indicates such a conclusion in the temporal domain makes no difference.

Pre-Juwaynian Critique of Analogy in Ibn Sīnā and Kalām

  • 55 For example: Abū l-Qāsim al-Anṣārī (2010), al-Ghunya fī ilm al-kalām, Cairo, Dār al-Salām, Muṣtafā (...)
  • 56 Cf. Abd al-Malik al-Juwaynī (1969), Shāmil, p. 660-665.
  • 57 Ibn Sīnā (1964), al-Shifā, Kitāb al-Qiyās, Cairo, al-Maktaba al-āmma, Saīd Zāyid (ed.), vol. 4, p (...)
  • 58 Ibn Sīnā (1964), al-Shifā, Kitāb al-Qiyās, vol. 5, p. 573-580

44Juwaynī’s theorization of analogical arguments into the four types discussed remained the standard characterization of inferences from present to absent, and they would continue to be discussed in chapters on God’s attributes.55 In what follows, I will investigate the reception of his formalization and critique, in addition to another line of critique originating in the tradition of falsafa, beginning with Fārābī and later Ibn Sīnā. Having said that, there is evidence to suggest that the lines are not so clearly demarcated. On the one hand, Juwaynī’s redundancy critique of the analogical argument is already present as early as Fārābī and Ibn Sīnā; while elements of Ibn Sīnā’s (d.427/1037) skeptical arguments are already present in figures like Ibn Fūrak (d.406/1015).56 It is plausible to think that the discussions around this inference were inter-disciplinary. Indeed, as we shall see, Ibn Sīnā does not present his critique of analogical arguments in Logic of the Shifa under the chapter of analogy (tamthīl), wherein he presents a routine Aristotelian account complete with its Greek examples of war between the Athenians and the Thebans.57 Instead, he discusses the kalām based inference in a subsequent chapter, where he combines internal kalām critiques in addition to the redundancy argument.58

  • 59 Abū Naṣr al-Farābī (2014), al-Manṭiq inda al-Farābī: Kitāb al-Qiyās, Beirut, Dār al-Shurūq, Rafīq (...)
  • 60 Abū Naṣr Farābī (2014), Kitāb al-Qiyās, p. 48.

45The logical critique of the analogy from present to absent is found at least as early as Fārābī in his Kitāb al-Qiyās.59 In his efforts to show the efficacy and utility of logic, Fārābī took the time to study the forms of argument found in Islamic law and kalām, and presented his analysis therein. The gist of Fārābī’s analysis is that that if we know with certainty, by non-inferential evidence or a distinct argument, that some judgement or property j is necessitated by a certain grounding property c, then knowledge of this proposition is sufficient, with no need for analogy between source cases and target cases.60 Arguments which succeed in this case may be analyzed as a first figure syllogism. But if we have no certainty regarding the cause of the judgement, then an inexhaustive list of resemblances between source and target will not be sufficient to justify the transfer of the judgement from source/known cases to target/unknown cases, and therefore, shall remain probabilistic and ampliative. In other words, either our evidence is sufficient to make the inference by virtue of the relation that holds between the grounded property and the grounding property, or it is insufficient. If it is sufficient, then any mention of the source domain is redundant; and if it is insufficient, then the inference cannot yield knowledge but only probable opinion.

  • 61 Cf. Ghazālī (2016), Miyār al-ilm, Beirut, Dār al-Minhāj, p. 190-196; and Fakhr al-Dīn al-Rāzī (20 (...)
  • 62 Cf. Schmitt, Jens Ole, (in this volume), “Preferring Formal Language over the Face? Avicenna on the (...)
  • 63 Ibn Sīnā (1964), Qiyās 9.24, p. 574.12-13.

46Ibn Sīnā develops Fārābī’s critique of the shāhid to ghāʾib arguments in Kitāb al-Qiyās 9.24. Ibn Sīnā dedicates a separate chapter for these kinds of inferences which he categorizes as inference from signs (dalīl or alāma), which he distinguishes from the immediately preceding chapter on analogy (Qiyās 9.23). Later authors starting with al-Ghazālī and others in the logical tradition such as Fakhr al-Dīn al-Rāzī, integrate the critique into a single chapter on analogy (tamthīl).61 This makes Ibn Sīnā’s treatment even more interesting: rather than treating this kind of argument as an analogy or a deduction or otherwise, he groups it together with the physiognomical syllogism.62 The commonality is the following: in both cases, the agent infers from a ‘sign’ to ‘the signified’. The observation of the origination of something, for example, is the sign pointing to its cause; while a certain facial feature is a sign pointing towards a deeper character trait. The only difference, it would appear, is that Ibn Sīnā states that such inferences are not syllogistic per se because they do not explicitly express the universal premise; if they did, he says, they would just be a syllogism.63

47After discussing the particular notions of dalīl and alāma¸ and how they can be analyzed in terms of the various syllogistic figures, Ibn Sīnā presents the following analysis of the inference from present to absent:

  • 64 Ibn Sīnā (1964), Qiyās, 9.24, p. 575.14-576.4.

A group of people who are called the ones who infer from the present/observable to the absent/unobservable seek to form all of their syllogisms out of signs (ꜥalāma), and they derive their universal premise from the major term and the sign (ꜥalāma). At times they do this through (i) positive induction (al-isitqrā al-mustawī), and sometimes (ii) through converse induction (al-istiqrā al-maꜥkūs), namely, that which is occurs through contrapositive of the sought conclusion. The first they call co-presence (ṭard), and the second they call co-absence (ꜥaks), and they call the sign (al-ꜥalāma) a grounding cause (ꜥilla). Then, when they seek to succeed in verifying [the cause], they enumerate the properties of something as in the case of the exemplar (al-mithāl), then they seek out the sign and cause by examining them one by one; or, they eliminate them one by one from being the sign, and they then judge that the remaining one is the sign or the remainders are signs, and by this they think they produced a proof.64

Ibn Sīnā argues that these inferences seek to form all of their inferences around signs. The sign, furthermore, is what they use to derive their universal premise from the major term and the sign (i.e., the middle term). For example, if the sign of being created is having a temporal origin (i.e., being ḥādith), then we combine ‘being originated’ with the major term which is signified by the sign, ‘having a cause,’ yielding our universal premise. For example:

  1. Zayd has a temporal origin (minor premise)

  2. Everything with a temporal origin has a cause (major premise)

Which yields ‘Zayd has a cause.’ Then Ibn Sīnā goes on to discuss the various methods they use to identify said causes or signs.

48According to Ibn Sīnā, the claim that they have two ways of identifying the ‘sign,’ either through a positive investigation, or through a negative investigation; a clear reference to collection and division (al-sabr wa-l-taqsīm). This form of inference is discussed both in Baghdādī and criticized by Juwaynī in the Burhān.

  • 65Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 286.

49Briefly put: Baghdādī identifies three types of ‘inference by division’ (al-istidlāl bil-taqsīm): (1) something is affirmed, and divides into two disjuncts, such that the negation of one disjunct implies the affirmation of the other, and vice versa (the implication here is that in such cases, the two disjuncts may be logically exhaustive); (2) something is affirmed and divides into multiple disjuncts, whereby one either investigates each possibility positively, seeking to prove one of the disjuncts and thereby negating the rest, or, investigating them negatively, refuting all and affirming the remaining disjunct (the implication, however, is that in the case of multiple disjuncts, the division is not logically exhaustive); and (3) something is affirmed and divides into two primary disjuncts (A and B), wherein one of those primary disjuncts then divides into multiple sub-disjuncts (e.g., B1, B2, B3). By refuting all of the sub-disjuncts, one refutes the primary disjunct (B), thereby proving the other disjunct (A). In this case, the two primary disjuncts are not logically exhaustive.65

  • 66Abd al-Malik al-Juwaynī (1981), Shāmil, p. 66-67.
  • 67Abd al-Malik al-Juwaynī (1981), Shāmil, p. 66-67.

50In the Shāmil, Juwaynī points out that inference by division was counted as a distinct type of inference by al-Bāqillānī, but he was contested on the following grounds: a division is not technically a type of evidence (dalīl). Division, rather, appears to be a tool by which we organize how we are to apply our inferences. In the case of division, the evidence is deployed in each individual disjunct, either to prove it or to disprove it, while the division does neither.66 Nevertheless, Juwaynī distinguishes two main types of inference by division: (i) binary divisions, such as ‘an existent is either eternal or originated’. It is true, of course, that ‘eternal’ and ‘originated’ are not contradictories, but they remain logically equivalent to a contradictory pair, because ‘eternal’ is defined as ‘having no beginning’ and ‘originated’ is defined as ‘having a beginning.’ And (ii) non-binary divisions which are not, and these are the ones which aim to identify the causes (illal) that are then deployed in our analogical arguments.67

  • 68Abd al-Malik al-Juwaynī (1978), Burhān, p. 131.
  • 69Abd al-Malik al-Juwaynī (1978), Burhān, p. 131.
  • 70Abd al-Malik al-Juwaynī (1978), Burhān, p. 157-158.

51In the Burhān, Juwaynī on the other hand, rejects both 2 and 3, claiming that these non-exhaustive disjunctions can never yield the certainty demanded of the theoretical sciences such as kalām.68 So long as these lists remain inductive, the possibility of another disjunct not considered by the inquirer remains, and therefore, the argument always remains doubtful. Nevertheless, Juwaynī notes that an exhaustive disjunct, which he describes as being ‘circumscribed between affirmation and negation,’ that is, between logical contradictories (or their logical equivalences as above), can form part of an effective proof.69 Indeed, he suggests that such logical disjunctions are essential to the entire enterprise of rational inquiry, which proceeds either by proving one and thereby refuting the other; or refuting one and thereby proving the other.70

52Returning to Ibn Sīnā: the context seems clear that what he means by ‘positive induction’ is what Baghdādī means in the second and third types identified above: namely, inductively deriving a list of possibilities and investigating each with the aim to prove one, and thereby refute the others. Likewise, the ‘converse’ or ‘negative’ investigation is to examine the disjuncts with the aim of refuting each one under examination, until one proves an isolated disjunct. Ibn Sīnā does not consider the case of an exhaustive disjunction.

  • 71 Ibn Sīna (1964), Qiyās p. 9.24, p. 576-577.
  • 72 Ibn Sīnā (1964), Qiyās p. 9.24, p. 576-577.

53At this point, he begins his critique through a series of skeptical arguments about their capacity to identify the cause.71 The objections are as follows: (i) perhaps the desired judgement (i.e., the grounded property) is not caused at all, and arises by virtue of the thing itself? If properties had a grounding property, this would entail an infinite regress. (ii) It is difficult to enumerate all the possible properties that ground the judgement. Perhaps one has failed to enumerate the actual cause in their induction. (iii) It is not necessary for the disjuncts to equal the number of properties. Perhaps, Ibn Sīnā argues, the cause is a combination of one or more properties. (iv) Lastly, Ibn Sīnā argues that perhaps the cause one identifies is too general, and that it too divides into subtypes which is in fact the true cause.72

  • 73 F. al-Dīn Rāzī (2002), Mulakhkhaṣ, p. 339-340.

54Fakhr al-Dīn al-Rāzī summarizes Ibn Sīnā’s critique in his Mulakhkhaṣ in the following way: Analogies can never yield certainty because (i) they rely on biconditional concomitance, i.e., co-presence and co-absence (ṭard and aks) to identify the cause, which is insufficient to determine what it truly is; and (ii) there remain the 4 avenues of skepticism as described by Ibn Sīnā in detail above.73

55Interestingly enough, these skeptical doubts appear to have been raised and debated heavily between Asharī and Mutazilī thinkers in the generation right before Ibn Sīnā. We can get a sense for these discussions from the reports by Juwaynī in the Shāmil. One of these is the chapter on whether the concomitance, i.e., co-presence and co-absence, of the two properties is sufficient for the identification of the cause (illa). After laying out the problem, and stating that all the ‘verifiers’ (al-muḥaqqiqūn) agreed that mere biconditional concomitance is insufficient for identifying the common cause, Juwaynī makes the following programmatic statement:

  • 74Abd al-Malik al-Juwaynī (1969), Shāmil, p. 661. The Arabic in the Nashshār edition reads: idhā lam (...)

If someone says: If mere biconditional concomitance (al-iṭṭirād wa l-inꜥikās) is not a sign for affirming something as the cause, then by what means do you know it? If knowledge obtains in a substrate, and the judgement holds for the substrate, then what guarantees that the knowledge is merely biconditionally concomitant to the judgement, but does not necessitate it? This is indeed a great matter of investigation regarding causes.74

  • 75 Cf. Young, Walter E. (2019), “Concomitance to Causation: Arguing Dawarān in the Proto-Ādāb al-baḥth(...)
  • 76Abd al-Malik al-Juwaynī (1969), Shāmil, p. 661.

Thus, there was a recognition of a general problem with the causes (i.e., middle terms) that play this essential role in all of their inferences: how to distinguish between two biconditionally concomitant properties, and between two properties one of which is causally dependent on the other.75 Juwaynī then cites Ibn Fūrak’s attempt to resolve the issue, with the aim of refuting him, and asserting the views of al-Bāqillānī (d.403/1012): 76

  • 77Abd al-Malik al-Juwaynī (1969), Shāmil, p. 661.

[Ibn Fūrak]: ‘So long as they are biconditionally concomitant, and no rational argument refutes its being a cause, then from this we know it is the true cause.’ [Ibn Fūrak] asserted this in his discussions by saying: ‘If all divisions are refuted, and one remains biconditionally concomitant, and no evidence stands to undermine it, then it is the cause.’ (Juwaynī) And this is extremely weak, and amounts to sufficing with a claim. This is because it is possible for the opponent to say: ‘How do you know you have exhausted all the divisions and refuted the rest, but what assures you – with all your efforts to avoid error – that you have not neglected a division and have not mentioned it? And if you mentioned it, that you would find it to be biconditionally concomitant and free of all rational objections?77

  • 78Abd al-Malik al-Juwaynī (1969), Shāmil, p. 661-663.
  • 79Abd al-Malik al-Juwaynī (1969), Shāmil, p. 664-665.

As is evident, Juwaynī’s skepticism regarding several non-exhaustive disjunctions appears to have existed long before he abandons the analogical argument by common cause. The rest of the discussion in the chapter continues to present objections to Ibn Fūrak’s views – and the views of Abū Isḥāq al-Isfarāyīnī – until finally presenting what he holds to be the appropriate alternative.78 Juwaynī opts for an additional condition to those mentioned by his predecessors, namely, positive evidence for the causal dependence of the judgement on the given property. This means that in addition to (i) biconditional concomitance, and (ii) the absence of contrary evidence and counterexamples, one must also have some (iii) positive evidence that the property in question is in fact the cause of the other. In the cases that Juwaynī cites, it seems quite evident. For example, Juwaynī states that if knowledge obtains in a subject, then we know with certainty that the subject is a knower. If anyone tries to claim that perhaps it is some property other than knowledge in the substrate that accounts for its being a knower, then they are engaging in sophistry, and have merely re-named knowledge by something else.79

56But what about other cases when the example is not so clear? Although Juwaynī here deals with many skeptical doubts regarding the methods to identify causes, he nevertheless defends the general view. The purpose of this discussion was merely to outline some of the precursors to his critique in the Burhān, and arguably, some of the cracks in these analogies by common cause were already beginning to show in his vigorous defense present in the Shāmil. Likewise, we have seen that Ibn Sīnā’s critique itself engages with these internal kalām discussions. To sum up: there are two general types of critique: (1) the redundancy argument that says if one actually succeeds in identifying the common element, then one has all they need to derive a universal premise and a middle term, rendering the source case irrelevant to the inference; or (2) one is unable to decisively identify the common element, due to the myriad ways one can raise doubts about it, thus rendering common-cause analogy ineffective in the rational sciences. Now I shall shift to the reception of these discussions.

The Early Post-Juwaynian Reception of the Critique of Analogy

  • 80 A. l-Qāsim Anṣārī (2010), al-Ghunyā, p. 520.
  • 81 A. l-Qāsim Anṣārī (2010), al-Ghunyā, p. 521. Barring teleportation and the interpenetration of like (...)
  • 82 A. l-Qāsim Anṣārī (2010), al-Ghunyā, p. 523-529.

57In the first generation of Juwaynī’s students, Abū l-Qāsim al-Anṣārī (d.512/1118) and Abū Ḥāmid al-Ghazālī (d.505/1111), we begin to see some divergences. After Anṣārī discusses Juwaynī’s standardized account of analogical arguments, he goes on to mention Juwaynī’s contention that if evidence is established for a given claim, then mentioning the source case has no impact.80 Anṣārī also mentions Juwaynī’s critique of non-exhaustive division, but qualifies this by stating that there are cases where the disjuncts are not logical contradictories of one another, but we can still know with certainty that the division is exhaustive relative to the case. For example, says Anṣārī, suppose we know that Zayd has exited a room with three exits, a, b, and c. We come to know that he did not leave through exit a or exit b, and therefore, we conclude that he must have left from exit c.81 Furthermore, Anṣārī continues to use the analogical arguments for the reality of God’s attributes in the same way that Juwaynī does in his Shāmil and Irshād.82 In other words, it does not appear to be the case that Anṣarī has accepted the redundancy argument made by Juwaynī.

58It may be worth revisiting the way in which known-source cases can play a role in the inference:

  1. In the temporal (source) domain, we have certain knowledge that ‘being a knower’ is caused by the property of ‘knowledge’.

  2. Causes necessitate their concomitant judgements.

  3. The Eternal is a knower (by an external argument).

  4. If the Eternal’s being a knower is not caused by the property of knowledge as it is in the source domain, then it is false that rational agents in the source domain need knowledge to cause their state of being a knower.

  5. But it is the case that rational agents in the source domain are knowers because of knowledge (by an external argument).

    • 83 A. l-Qāsim Anṣārī (2010), Ghunyā, p. 519-520.

    Therefore: the judgment that the Eternal is a knower must be grounded in the property of knowledge.83

  • 84Abd al-Karīm Shahrastānī (1932) Nihāya, p. 182.

In other words, the way that Anṣārī preserves the relevance of the source domain is by making the same claim that Juwaynī had made in the Shāmil, namely, that if we allow for the specification of this cause to a single domain, such that in some domains it is effective and in other domains it is not, then it cannot be a cause at all. But we know with certainty that it is a cause, and therefore, if we do not extend that judgement to the target domain, we must also reject our knowledge of the source domain. Anṣarī’s most famous student, al-Shahrastānī (d.548/1153), reiterates this line of reasoning with vigour in his Nihāyat al-iqdām.84

  • 85 A. Ḥāmid Al-Ghazālī (2016), Miyār al-ilm, p. 190.

59Ghazālī seems to have gone a different route. In his Miyār al-ilm, Ghazālī identifies the inference from present to absent as a form of tamthīl, the technical term in Aristotelian logical writings for analogical arguments.85 There are reasons to doubt this categorization by Ghazālī. Recall that Ibn Sīnā does not categorize this kind of inference as a form of tamthīl, and rather affords it its own separate discussion. Secondly, in our analysis of Baghdādī and Juwaynī above, although these arguments have an analogical element, they resolve to what is a deduction with the same power contained in a first figure syllogism. Nevertheless, Ghazālī makes the equation and reiterates the redundancy argument:

An example is to say: the heavens began to exist, because it is a body, and therefore it began to exist, in analogy with plants and animals, for these bodies which we observe began to exist.

But this is invalid so long as it is not proven that the plants began to exist because they are bodies, and that its corporeality is the middle term for its origination (ḥudūth). But if that is proven, then you know that animals are originated because all bodies are originated, and that is a universal judgment. From this a syllogism in the first figure is formed, which is ‘the heavens are a body’ and ‘every body is originated,’ yielding ‘the heavens are originated.’

  • 86 A. Ḥāmid Ghazālī (2016), Miyār, p. 190.

Thus: the transfer of the judgement from the universal to the particular comes under this, and it is valid; and the effect of the particular observable case is nullified, and so the mentioning of the animal case is extraneous to the argument.86

Thus, Ghazālī clarifies the redundancy in logical terms: if we know the cause, that is, the middle term, which explains why a certain judgement holds, then we have a universal judgement. But if we have a universal judgement, then we can just use this as the major premise of a syllogism in the first figure, and the source case no longer serves any purpose; its relation to the major premise is just the same as the target case.

  • 87 A. Ḥāmid Ghazālī (2016), Miyār, p. 193.
  • 88 A. Ḥāmid Ghazālī (2016), Miyār, p. 193.

60Ghazālī then presents himself with the following question: ‘if this inference is redundant and what you say is certain, then how can we believe that the theologians (mutakallimūn) in all their numbers and the precision of their intellects had missed this?’87 Ghazālī presents two possible answers. (1) If the argument may be analyzed as a valid deductive argument, then the reason why the mutakallimūn mention the source case is not to draw an analogy, but in order to caution (tanbīh) their interlocutor by drawing attention to the universal premise implied in an uncontroversial case.88 In that sense, it is not part of the general inference but is rather external, supporting evidence in dialectical contexts.

61Ghazālī’s critique, although it corresponds to the gist of Juwaynī’s in its meaning, still appears to be coming from the tradition of logic as opposed to the internal critique of kalām. Nevertheless, we can find grounds among early kalām scholars for his claim that mentioning the source case can serve a dialectical or communicative purpose when presenting a deductive argument. Ibn Fūrak cites Asharī’s position as stating the following:

  • 89 A. B. M. b. al-Ḥasan Ibn Fūrak (1987), Mujarrad, p. 187.

We do not say that whatever is absent from us is living, knowing, and powerful in analogy (qiyāsan) to the fact that we have not observed an agent that was not living, knowing or powerful; and whoever holds that is the case is in error. Rather, we say that our knowledge of the Eternal as living, knowing, and powerful, is due to the appearance of His precise and ordered creative acts. That is because if it were possible for the World to come into existence in this precise manner from a non-knowing, non-living, non-powerful being, then the appearance of accidents from one who is not knowing, non-living, nor powerful, would be even more plausible. Thus, the creative acts of the Eternal indicate that He is living, knowing, and powerful, just as the acts of human agents, if they are precise and orderly, indicate that they are living, powerful, and knowing, because the manner of knowing that a knower amongst us is knowing, powerful, and living, is by inference, not by observation.89

In this early example, just as we saw more clearly articulated by Juwaynī in the Shāmil, there is a clear rejection of an arbitrary generalization from mere observation. It is not the case that we infer from the fact that we have not observed any agent that was not living, knowing, and powerful, that all agents must be so. But rather, just as we infer from wise or precise acts in the case of human agents, we infer from wise and precise acts in the eternal case. In such a case it is plausible for Ghazālī to infer that mentioning the known source case may support one’s argument because it helps to illustrate the universal principle. The difference between Ghazālī and his predecessors, therefore, appears to be whether or not it is necessary to mention the source case at all; Juwaynī and Ghazālī appear to hold that it is extraneous to the argument.

  • 90 A. Ḥāmid Ghazālī (2016), Miyār, p. 194-196.

62(2) Ghazālī goes on: if the argument cannot be analzyed to deductive proof, then the other possibility is that the thinkers in question were simply mistaken, either due to an erroneous abstraction from a single case or due to an incomplete induction thought to be exhaustive; but this should not worry us, because we should know truth by reasons and not by men.90 In other words, unlike his contemporary and fellow student al-Anṣārī, Ghazālī accepts the critiques found in Ibn Sīnā and Juwaynī in the Burhān, and holds that all such inferences should be abandoned in the rational sciences.

The Later Reception of the Reception of the Critique of Analogy

63Several commentaries were written on Juwaynī’s Burhān and Irshād in the 6th/12th-7th/13th centuries. It was during this period, after the first generation of his students Anṣārī and Ghazālī, that two interrelated disputes took place: one on abstract modes and another on inferences from the present to the absent. With respect to the inference from present to absent, there is agreement that the argument succeeds; there is also agreement that, when an analogical inference succeeds, it can be analyzed as being deductively valid. The disagreement, however, appears to be on whether or not the source case plays a role in the inference or not. Fārābī, Juwaynī, and Ghazālī hold to the redundancy view; while other figures in the reception of the critique maintain that the source case still plays a critical role in the argument. We saw already that for Anṣārī, the source case remains relevant because if we do not transfer the judgement to the target case, we undermine the source case. In this way, there is a way of keeping the source case relevant to the overall inference.

  • 91Abd al-Karīm Shahrastānī (1932), Nihāya, p. 182-183.

64The Asharī mutakallim and student of Anṣarī, al-Shahrastānī, appears to approve of Juwaynī’s objections to analogy. In his Nihāyat al-iqdām, Shahrastānī begins the discussion by mentioning the nature of the analogical inference and the four types and defends this inference aptly.91 Nevertheless, he then poses a plausible objection against himself. If one agrees that the dependent property, such as being a knower, is grounded in the concrete property of knowledge in the observable domain, then in order for them to extend that judgement to the divine target case, then they need to show that the dependent property is equivalent in both cases. However, God’s being a knower is clearly different – categorically – from the case of human knowers. But since the dependent property is distinct in several crucial ways, it is doubtful whether it should also hold for eternal, unobserved cases. Thus, Sharhastānī argues:

  • 92Abd al-Karīm Shahrastānī (1932), Nihāya, p. 186.

There is no need to bother with a common element between present and absent by virtue of grounding cause and dependent property, but rather: if the evidence indicates the conclusion in the unobservable in that He is knowing with knowledge, powerful with power, then that evidence is sufficiently independent and is not in need of noticing the source (shāhid, lit. ‘present’) case.92

65Other commentators on Juwaynī’s argument seem to disagree. The Sicilian Mālikī jurist al-Māzarī (d.536/1141) holds that Juwaynī’s objection is vacuous and that the disagreement between the two groups is merely verbal:

  • 93 M. b. Alī Al-Māzarī (2000), Īḍāḥ al-maḥṣūl min Burhān al-uṣūl, Tunis, Dār al-Gharb al-Islāmī, Amm (...)

It is said to Abū l-Maꜥālī: If we suppose that the evidence established for us applies in the absent and the present domains, and its conception in the absent is the same as its conception in the present, then: do you deny the equivalence between the absent and the present with respect to what the evidence implies? He must answer by stating that he does not deny this, and far be it from him to disagree with this, for there are only two options…So, what remains is only a dispute over the expression, and naming this an analogy when it is in fact known from the same evidence.93

Māzarī’s point appears to be this: either Juwaynī agrees that the evidence (i.e., precise orderly actions) indicates the existence of knowledge, or it does not. If it does, then he must accept that the evidence signifies the same thing in both the temporal and the eternal cases, namely, some minimal conception of knowledge or being a knower. Now, Māzarī denies, just as earlier thinkers did, that just because the evidence holds for the temporal case, that we can automatically transfer the judgement to all cases. Māzarī is not exactly disagreeing with Juwaynī here; he seems rather to be stating that Juwaynī has said nothing new.

66Another important figure from the 6th century was Taqī al-Dīn al-Muqtaraḥ (d.612/1215), who wrote commentaries on Juwaynī’s Irshād and Burhān. Muqtarāḥ mentions Juwaynī’s redundancy argument against the inference from present to absent, then makes the following reply:

  • 94 Taqī al-Dīn al-Muqtaraḥ (2014), Sharḥ al-Irshād, Tatouan, al-Rābia al-Muammadiyya, vol. 1, p. 314 (...)

This is rejected insofar as the evidence for the evidence (dalīl al-dalīl) may be specific to the present domain, and likewise, the evidence for the causal relation (al-taꜥlīl), and therefore, it stands in need of the present domain because of that. Furthermore, the absence of evidence of a causal relation in the absent domain does not indicate the negation of causation therein, for evidence does not imply conversely.94

Muqtaraḥ here argues that while it may be true that, once we know what the evidence for a certain conclusion is, we do not need to mention the source case, nevertheless, (i) the evidence for the evidence and (ii) the evidence for the causal relation or the concomitance between two items may not be knowable except through an analysis of the source domain. If we take the knowledge case as an example, then it may be that we would never know that precise orderly acts signify knowledge if we were not made privy to it through our inferences in the temporal source domain. Likewise, we would not have discovered that ‘being a knower’ is a judgement or an abstract mode necessitated by the property of knowledge unless we had recourse to the source domain. In other words, we discover that (a) precise acts indicate knowledge and (b) being a knower is concomitant to having knowledge, through the evidence we find in the source domain. On this view then, the inference from present to absent becomes a composite inference:

Muqtaraḥ’s inference: (i) We inquire into the temporal/present domain and deduce that evidence A indicates the judgement J, and the judgement J necessarily implies property C; (ii) then we discover new evidence A1, which indicates judgement J1 in the absent/eternal domain; and from (i) we further infer that since J implies C by necessity in the temporal/present domain, then J1 implies C1 by necessity in the eternal/absent domain.

  • 95 Taqī al-Dīn al-Muqtaraḥ (2014), Sharḥ al-Irshād, vol. 1, p. 314.
  • 96 Taqī al-Dīn al-Muqtaraḥ (2014), Sharḥ al-Irshād, vol. 1, p. 286, p. 326.
  • 97 Taqī al-Dīn al-Muqtaraḥ (2010), Sharḥ al-Burhāniyya, Beirut, Maktabat al-Maārif, Nizār Ḥammādī (ed (...)

Nevertheless, al-Muqtaraḥ agrees with Juwaynī’s position that the denial of the theory of aḥwāl eliminates the possibility of the inference from present to absent at least in the case of the common cause.95 He also appears to hold that at least in theological questions, mentioning the source or present case does not add anything to the inference.96 Indeed, in a later work, he reiterates the claims that (i) inference by common cause is dependent on the theory of abstract modes or (ii) that the present and absent fall under a single concept; but the theory of modes is false, and God’s attributes are utterly distinct from all human attributes, and, therefore, do not fall under the same concept. Therefore, it suffices for us, in the inference, to argue from God’s acts instead, without recourse to analogy.97

67A similar point regarding the non-redundancy of the source case is made by Muqtaraḥ’s contemporary al-Abyārī (d.618/1221) in his Sharḥ al-Burhān:

  • 98 Al-Abyārī (2013), al-Taḥqīq wa l-bayān fī sharḥ al-Burhān, Kuwait, Dār al-iyāʾ, vol. 1, p. 433.

The statement of the Imām (i.e., al-Juwaynī): that ‘the evidence established in the present domain, if it is established in the absent domain, then it suffices and there is no need to use the present case as evidence.’ This is true, except that the mind does not recognize the evidence except after reflecting on the present domain.98

  • 99 Abyārī (2013), Sharḥ al-Burhān, vol. 1, p. 433.

The point is reminiscent of Ghazālī’s statement that the source case helps one recognize the more general principle, but the inference as such remains independent of the particular cases of ‘animal’ or ‘plant’ and so on. But Abyārī follows up by stating that when one reflects on the concept of ‘being a knower’ in the present domain, one concludes that it cannot be reducible to the pure quiddity of the subject; rather, it must be grounded in an additional property, namely, ‘knowledge.’99

The Case of Fakhr al-Dīn al-Rāzī (d.606/1210)

  • 100 A. Shihadeh (2005), “From Al-Ghazālī to Al-Rāzī,” in Arabic Sciences and Philosophy, p. 165-168.
  • 101 K. El-Rouayheb (2016), Review of The First Islamic Reviver by Kenneth Garden and Al-Ghazālī’s Moder (...)

68In his classic article “From al-Ghazālī to al-Rāzī,” Ayman Shihadeh argued that the impetus for the transition of kalām from its classical phase to its post-classical phase should be attributed to Rāzī and not Ghazālī, for two reasons: (i) Ghazālī’s kalām work relied on classical modes of argument which were invalid and dialectical; and (ii) Rāzī’s work uses logical methods of demonstration and rejects the classical modes of inference.100 El-Rouayheb has sufficiently refuted the first claim, and in what follows, I will present some reasons for doubting the second claim.101 That is not to say that Rāzī is not the most important figure in the post-classical tradition: the fact that he continues to be labeled “al-Imām” in that tradition – replacing al-Juwaynī – is not by accident; but some of the reasons Shihadeh enlists as evidence for the claim are dubious.

  • 102 A. Shihadeh (2005), “From Al-Ghazālī to Al-Rāzī”, Arabic Sciences and Philosophy, p. 165-168.

69In particular, Shihadeh cites Rāzī’s critique of the following inferences: (i) the argument from the absence of evidence; (ii) analogical arguments, which includes a critique of the two methods of discovering causes: (a) concomitance and (b) incomplete divisions; (iii) ‘the argument ad hominem’ which Rāzī identifies as a form of analogy; (iv) reliance on scriptural evidence in an ad hoc manner and the need to divide objects of knowledge into (a) what can only be known by reason, (b) what can only be known by revelation, and (c) what can be known from both.102

  • 103 A. B. Bāqillānī (1998), al-Taqrīb wa l-Irshād, Beirut, Al-Resalah Publishers, Abd al-amīd Abū Zay (...)
  • 104 A. Shihadeh (2013), “The Argument from Ignorance and Its Critics,” Arabic Sciences and Philosophy, (...)

70The main issue with citing these as evidence is that all the items listed are found in Juwaynī’s Burhān and Shāmil in almost the same order; and the division of theological theses into the three categories of reason alone, revelation alone, and both, is found at least as early as Bāqillānī.103 They originate in the inference types mentioned by Baghdādī in his Iyār and critiqued by Juwaynī in his Burhān. In other words, the dispute over these forms was an internal carry-over from the classical tradition and is not distinctive of the transition to the post-classical phase. Rāzī, whose father was the student of Juwaynī’s student Anṣārī, was no doubt well aware of these critiques from a very young age, and he is simply expanding on a discussion that – as we have seen above – was already taking place. In a later article, Shihadeh recognizes the influential nature of the Burhān, but does not recognize that the modes of inferences critiqued by Rāzī were already critiqued long before the integration of logic, and, therefore, cannot serve as evidence to support Rāzī’s role in moving kalām into its post-classical phase.104

  • 105 F. al-Dīn Rāzī (2015), Nihāyat al-uqūl, Beirut, Dār al-Dhakhāʾir, Saīd Fūdā (ed.), vol. 1, p. 136 (...)
  • 106 F. al-Dīn Rāzī (2015), Nihāya, vol. 1, p. 137.
  • 107 F. al-Dīn Rāzī (2015), Nihāya, vol. 2, p. 235.

71Indeed, when Rāzī discusses the question of arguments from present to absent, he combines elements of all past critiques, but he nevertheless follows the same line and uses some of the same justifications and examples; in his discussion of cause-and-effect inferences, he notes, like Juwaynī, ‘Being moving is identical to motion according to us, for we do not hold the theory of cause and effect.’105 In his discussion of an incomplete division, Rāzī uses the same example as Juwaynī does, namely the enabling cause of visibility, in relation to discussions of the Beatific vision.106 Rāzī also does not shy away from affirming the equivalence between the ‘present/temporal’ and the ‘absent/eternal’ domains in his discussion of the reality of the Divine Attributes: “Furthermore, the reality of ‘being a knower’ insofar as it is ‘being a knower’ does not differ in the present and absent domains, and thus, it is necessary to judge that God’s being a knower is a predicate that is additional to His Essence.”107 That is, when one equates the two properties of ‘being a knower’ and ‘knowledge,’ as held by Asharī, Juwaynī, and all others who rejected the theory of aḥwāl, then one no longer stands in need of the further inference from the judgement to the property, as we saw above. Rāzī here is operating squarely within the parameters of the Juwaynian branch of Asharism, of which he is an heir. Much more could be said regarding Fakhr al-Dīn’s expansion of Juwaynī’s criticism of Baghdādī, but for the purposes of this paper, my intent was only to show that this discussion had little to do with the adoption of Avicennan logic.

The Afterlife of Analogical Arguments in Kalām

72We have seen that the jury was split in the post-Juwaynian period between figures who accepted Juwaynī’s critique and those who did not. Figures like al-Muqtaraḥ, for example, believed that the known-source cases still played an essential role in the inference, and that the argument from present to absent could not be understood without them. Nevertheless, the oft-heard claims regarding the integration of logic into the kalām tradition make it easy to accept that such inferences must have faded away with the new tools, and that the old ‘dialectical’ ways of classical kalām were readily replaced. Interestingly enough, in the post-Rāzīan period, we still do not see the abandonment of such inferences; in fact, a significant number of later scholars who are fully engaged in the logical tradition do not abandon the use of ‘analogical’ arguments, nor do they abandon the possibility of analogies yielding certainty. In what remains of this paper, I want to canvas some of these views and point to areas that still require further research. Here I would like to present the views of two major figures active in Khorasan and Central Asia: Ṣadr al-Sharīa al-Maḥbūbī (d.747/1346) in his Tadīl al-manṭiq, and Sad al-Dīn al-Taftazānī (d.793/1390) in his Sharḥ al-Maqāṣid, with some of the latter’s commentators.

Ṣadr al-Sharīꜥa on Analogy and Experimental Knowledge

73As is usual in the logical tradition after Ibn Sīnā, Ṣadr al-Sharīa’s ends his advanced work on logic by discussing the types of propositions needed in constructing proofs. But somewhat unusually in works of logic, Ṣadr al-Sharīa then discusses induction and analogy as cases which depend on observation but only yield plausible opinion (ẓann), though at times they may yield certainty:

  • 108 Ṣadr al-Sharīa al-Maḥbūbī, Sharḥ Tadīl al-ulūm, Topkapı Sarayı Müzesi Emanet Hazinesi, ms. 1669, (...)

[Induction and analogy] may also yield certainty, as in the case where induction is complete, for the certainty yielded in sense perception is based on the absence of the causes of errors in sense perception, and that is not known except through induction; and in the case of analogy where knowledge of the cause is certain, and the certainty of knowledge by experience is based on this.108

Against some classical logicians who regularly denied that induction was ever complete, Ṣadr al-Sharīa provides us with a concrete example which we all ought to accept, namely, that our knowledge via sense perception is certain if and only if all the causes of error are absent, and we only know the causes of error through induction. Thus, if we concede that we have not exhausted all the sources of error, the doubt we have about our induction would extend to our sense perception. But since we do not have such doubts, we are committed to the fact that our induction of the possible causes of error is at least exhaustive enough to yield certainty.

74As for analogical reasoning, Ṣadr al-Sharīa continues with an explanation on how it forms the basis for another category of certain knowledge, namely experimental knowledge (al-mujarrabāt):

  • 109 Ṣadr al-Sharīa, Sharḥ Tadīl al-ulūm, fol. 108.

For when you try scammony, then you see another instance of scammony like the one you tried, you achieve certainty that it is a purgative, and that is not based on other than analogy, for if it were not for your judgement that this [scammony] is like that which was tried, and that the cause of purgativity was shared between them, then you would not have certainty that [scammony] was a purgative. Therefore, it becomes known that the certainty of experimental knowledge is based on analogical reason where [knowledge] of the cause is certain, and God knows best.109

Against the Avicennian orthodoxy that analogical reasoning could never reach certainty in practice, Ṣadr al-Sharīa shows rather convincingly that it seems reasonable for them to accept it; at least on the grounds that most logicians accepted that knowledge from experience yielded certainty. The key insight presented by Ṣadr al-Sharīa, therefore, is showing how knowledge from experience is in fact grounded in analogy; this should not be understated, for such trials and experiments are the basis for much of the natural sciences and pharmacology. Although beyond the scope of this research, Ibn Sīnā might actually agree here: in the Burhān of the Shifā, he subjects experience (tajriba) to the same kind of critiques he subjects the inferences from present to absent, except that in the case of experience, he determines that it yields a type of tentative or conditional necessity, while in the case of analogy he claims that it does not yield certainty at all.

Saꜥd al-Dīn al-Taftāzānī and Gelenbevī on Analogical Arguments

  • 110 Sad al-Dīn al-Taftāzānī (2012), Sharḥ al-Shamsiyya, ed. Jādullah Bassām, Amman, Dār al-Nūr, p. 370 (...)
  • 111 Cf. Sad al-Dīn al-Taftāzānī, in Young, “Concomitance to Causation,” p. 265.
  • 112 Sad al-Dīn al-Taftāzānī (2012), Sharḥ al-Shamsiyya, p. 372.
  • 113 Sad al-Dīn al-Taftāzānī (2012), Sharḥ al-Shamsiyya, p. 372.

75In his commentary on the Shamsiyya, Taftāzānī affirms the standard view that analogical arguments, including the classical kalām inference from present to absent, do not yield certainty.110 After presenting the standard logical formalization of the argument, he goes through the standard methods of ascertaining the cause followed by the skeptical arguments that show that grounds for doubt remain.111 On the assumption that we do know the cause and that it transfers, then mentioning the source case becomes redundant, for we have instead a universal major premise.112 Taftāzānī ends the discussion by claiming there is no disagreement among logicians that induction and analogy do not yield certainty.113

  • 114 Sad al-Dīn al-Taftāzānī (1860), Sharḥ al-Maqāṣid, vol. 2, p. 54-55.

76Nevertheless, in his later Sharḥ al-Maqāṣid, Taftāzānī repeats Juwaynī’s formalization of analogical arguments into four different types (cause, reality, condition, and evidence) before presenting an analogy as his first argument for the reality of God’s attributes.114 He does not follow up with a qualification that indicates he does not believe the argument holds or that such analogies are doomed to fail. Indeed, even the well-received 18th Century Ottoman logician Ismāīl Gelenbevī (d.1205/1791), repeats Taftāzānī’s defense of the analogical argument in his gloss on Dawānī’s (d.908/1502) Sharḥ al-Aqāʾid al-Aḍudiyya. Dawānī writes the following:

  • 115 Jalāl al-Dīn al-Dawānī (1898), Sharḥ al-Aqāʾid al-Aḍudiyya, vol. 1, p. 273.

Being a knower is caused by the subsistence of knowledge in it in the present domain, therefore, the case is the same in the absent; and extend this to the rest of the attributes…and the weakness of this inference is evident, for analogy between the absent and the present is a juridical analogy with an [invalidating] difference. Do you not see that power may be annihilated in the present domain, and it may increase and decrease, nor is it effective according to al-Ashꜥarī and his followers in that? While in the absent (eternal) domain it is distinct from all that.115

  • 116 I. Gelenbevī (1898), Sharḥ al-Aqāʾid al-Aḍudiyya, vol. 1, p. 272-273.

In his gloss, Gelenbevī cites Juwaynī’s argument as it is presented by Taftāzānī; he then presents Jurjānī’s rejection of the argument in his Sharḥ al-Mawāqif, whose arguments against the efficacy of the analogy are reproduced by Dawānī in the citation above.116 Gelenbevī then replies:

  • 117 I. Gelenbevī (1898), Sharḥ al-Aqāʾid al-Aḍudiyya, vol. 1, p. 273; cf. Taftāzānī (1860), Sharḥ al- (...)

As for denying that this is an invalid analogy, this has been answered by [Taftāzānī] by stating that the differences they have mentioned do not undermine the validity of the analogy, because these judgements do not differ, nor do the attributes, with respect to our aim here. For knowledge entails that a person is a knower insofar as it is knowledge, not insofar as it is incipient or eternal, or an accident or otherwise.117

This is the first part of the argument responding to Dawānī’s rejection of the analogical argument. Basically, the ways in which created knowledge is different from eternal knowledge are irrelevant to the ontological dependence of ‘being a knower’ and the grounding property of ‘knowledge.’ Gelenbevī then adds a further argument to show that analogical arguments, when they meet the conditions required, do yield certainty:

  • 118 I. Gelenbevī (1898), Sharḥ al-Aqāʾid al-Aḍudiyya, vol. 1, p. 273.

As for the claim that analogy cannot yield certainty, this is refuted by the scholars of legal principles who affirmed that analogies where the cause is explicitly known in the source case yield certainty, and since the cause of being a knower in the present domain and being powerful, for example, is knowledge and power without any doubt, the argument is equivalent to an analogy where the cause is explicitly stated, for the need for explicit [mention of the cause] in revealed evidence is in order to designate the cause with certainty, so understand this.118

Gelenbevī first argues that it is mere obstinance (mukābara) to deny that ‘being a knower’ is grounded in having the property of knowledge. Furthermore, the view that analogy cannot yield certainty stands in stark contradiction with the scholars of legal principles (uṣūl al-fiqh). They held that analogical arguments did in fact yield certainty when the common cause or ratio was stated explicitly by the Lawgiver. In such cases, one needed merely to identify the presence of the cause in a new case, and apply the judgement. It is only when the legal cause is derived theoretically, through the inductive methods discussed earlier, that the analogy is considered only probable. Thus, if the cause by which the judgement is transferred to the target case is known with certainty, then the argument yields certainty. Since Gelenbevī draws his inspiration from the science of legal principles where the source case is essential for the inference to the target case, it is plausible to suppose that Gelenbevī here does not think that mention of the source case is redundant. Thus, we see that even by the 18th century, the efficacy of the classical analogical inference from the present domain to the absent domain remains in place.

77As a final note to this section, Taftāzānī’s great grandson and contemporary of Dawānī Aḥmad b. Yaḥyā al-Harawī (d.916/1511), presents a compelling counter to the redundancy argument against analogical arguments. In his commentary on his great grandfather’s widely studied primer in logic, Tahdhīb al-manṭiq, Harawī presents the usual skeptical arguments against certain knowledge of the cause, then writes the following:

But know that it is possible to know the causality of the common element in some cases by an external proof, such as the causality of contingency in the need for a cause, in which case, analogical arguments yield certainty.

  • 119 A. b. Y. al-Harawī (2019), Sharḥ Tahdhīb al-manṭiq, Amman, Dār al-Nūr, Abd al-Ḥamīd al-Turkmānī (e (...)

Let it not be said: there is no need for the analogy and its premises, for it is sufficient to consider the form of the syllogism; for we say: if the opponent concedes the resemblance judgement but not the minor (i.e., the target), then presenting an analogy in refuting him (ilzāmihī wa ifḥāmihi) is even more forcibly expressed; furthermore, what the objector here says is equally directed against him in cases where the causality of the common element is strictly probable, for it is then possible to communicate the aim by means of a rhetorical argument instead of the analogy, so consider this.119

Harawī acknowledges that an external argument can yield certainty regarding the common cause. But then, as we saw in earlier authors, one may be satisfied with the syllogistic inference instead and dispense with the analogy, on the assumption that the source case is redundant. Harawī denies this for two reasons. The first is that analogies can have additional effectiveness in dialectical contexts. For example, in cases where an interlocutor does not concede the target case but concedes the resemblance between source and target, then the argument by analogy becomes much more compelling. That is, in a dialectical context where the opponent (i) concedes the judgement for the source case, and (ii) concedes that the source and the target share in the common cause, but seeks to deny (iii) the target judgement, then the use of an analogy becomes much more effective in forcing the opponent to concede. Indeed, this is the very same logic behind Baghdādī’s inferences in his Jadal discussed earlier.

78The second argument Harawī raises is that the certainty or lack thereof appears to be irrelevant. That is, dialectically speaking, if the opponent claims the analogy is useful just in case the common cause is probable, then how does the interlocutor deal with the following analogous objection: why bother mentioning the source case at all, and why not suffice with a rhetorical argument instead? But if the opponent sees a utility in the analogy that is not present in the rhetorical argument, then whatever that utility is, it should not have anything to do with the relative strengths of the two forms; and therefore, the same kind of utility can exist when the deduction is decisive.

Conclusion

79The classical kalām tradition deployed a form of argument known as the analogy from the present to the absent. On closer analysis, it does not in fact differ from syllogistic arguments in strength: if the common element between source and target is known with certainty, then the conclusion follows from the premises by necessity, just the same as it would in a syllogistic deduction.

80Asharī and Mutazilī thinkers distinguished between arbitrary forms of the inference and justified forms. As early as Farābī, however, there was a recognition that there are three forms in which the inference might take place: (i) if it is grounded in a necessary concomitance that applies equally to the source and the target case, we may simply dispense with the analogy and deploy a deductive inference; (ii) if the common cause is merely probable, then the inference is probable and cannot yield certainty; (iii) arbitrary analogies are those which rely merely on the observation of the case in the source target, without an attempt at identifying a further common cause.

81Juwaynī built on the work of his predecessors in formalizing the inference into the form accepted by the later tradition. He divided the inference into four types with respect to the common element in source and target domains: (i) common cause; (ii) common reality; (iii) common condition; and (iv) common evidence. In his later work al-Burhān, however, Juwaynī presents a critique of all these forms of inference, alongside others, as they are found in the Jadal of his predecessor al-Baghdādī. With respect to the analogical arguments in theology, Juwaynī considered the mention of the source case to be redundant and to be dispensed with; he states that analogies are categorically to be rejected in rational questions.

82Around the time of al-Juwaynī’s death, the Avicennian tradition of logic began to be integrated in the kalām tradition. Ghazālī follows this tradition in judging that ‘analogical’ arguments, if they succeed at all, are in fact deductive arguments, and the mentioning of the source case is redundant except in the case where an interlocutor has trouble grasping the universal premise. Juwaynī’s student Anṣārī, however, continues to deploy the inference in his work, implying that he does not concede that mentioning the source case is extraneous. His student al-Shahrastānī likewise uses the inference but somewhat tentatively, opting for other arguments in his work.

83Following generations saw the continuity of this dispute. Fakhr al-Dīn al-Rāzī follows Juwaynī in the Burhān and develops his critiques of Baghdādī. I pointed out that, contra Shihadeh, Rāzī’s critique of classical kalām methods in his Nihāyat al-uqūl is an internal critique that follows Juwaynī very closely. Nevertheless, some of Rāzī’s contemporaries, such as al-Muqtaraḥ and al-Abyārī, do not accept the redundancy claim. In particular, al-Muqtaraḥ holds that it is true that it may be true that a universal premise suffices one to form a deductive inference in the target domain without reference to the source domain, but the evidence of the evidence still remains necessarily grounded in the source domain. Therefore, it remains essential to the overall inference.

84Al-Muqtaraḥ was very influential in the North African tradition, and his defence of the inference remained in place even with the integration of logic. His student Ibn al-Tilimsānī likewise deploys the method in his commentary on Rāzī’s Maālim, and it is repeated by figures like Saīd al-Uqbānī in his Sharḥ al-Burhāniyya and al-Sanūsī in his Sharḥ al-Sanūsiyyā al-Kubrā, along with his commentators, such as al-Yūsī.

85In the Eastern Islamic tradition, I looked at a number of post-classical figures who also affirmed the inference from present to absent, and, more generally, defended both the indispensability of analogical arguments for some categories of definitive knowledge, and their indispensable utility in dialectic. Ṣadr al-Sharīa demonstrates convincingly, against much of the logical tradition before him, that our knowledge from trial and experiment is grounded in analogical reasoning. Taftāzānī, similarly, deploys Juwaynī’s classical formulation of analogical arguments in his large kalām summa Sharḥ al-Maqāṣid; his defence of the inference is picked up by later Ottoman thinkers such as Gelenbevī in order to deflect the skeptical claims of others such as Dawānī. Gelenbevī adds a reference to the science of legal principles in order to justify that if we have an external argument to demonstrate the common element, then the analogy yields certainty. Taftāzānī’s great grandson al-Harawī likewise provides a defence of the certainty of analogical arguments and rejects the redundancy claim for two reasons: (i) that an analogical argument, when it is deductively valid, can be more effective in refuting an opponent than its syllogistic form; (ii) the same argument can be made against probable analogical arguments by stating they can be dispensed with for a rhetorical argument with non-certain premises. Harawī therefore opens up an opportunity for further reflection on the need for and utility of analogical arguments beyond the question of certain knowledge.

Haut de page

Bibliographie

al-Anṣārī, A. l-Qāsim (2010), Al-Ghunya fī ilm al-kalām, M. Abd al-Hādī, Ed.; Vol. 2, Dār al-Salām.

al-Baghdādī, Abd al-Qāhir (2019), Iyār al-naẓar fī ilm al-jadal, A. Arrūbī, Ed., Asfār.

al-Bāqillānī, A. B. (1998), Al-Taqrīb wa-l-Irshād, Abd al-Ḥamīd Abū Zayd, Ed.; Second Edition, Vol. 1-3, Al-Resalah Publishers.

al-Dawānī, J. al-Dīn (1890), Sharḥ al-Aqāʾid al-Aḍudiyya (Vol. 2), Der Saadet.

al-Farābī, A. N. (2014), Al-Manṭiq inda al-Farābī, R. Ajam, Ed., Dār al-Shurūq.

al-Ghazālī, A. Ḥāmid (2016), Miyār al-ilm, Dār al-Minhāj.

al-Harawī, A. b. Y. (2019), Sharḥ Tahdhīb al-manṭiq, Abd al-Ḥamīd al-Turkmānī, Ed., Dār al-Nūr.

al-Juwaynī, Abd al-Malik (1969), Al-Shāmil fī uṣūl al-dīn, Alī Sāmī al-Nashshār, Ed., Manshaʾat al-Maārif.

al-Juwaynī, Abd al-Malik (1978), Al-Burhān fī uṣūl al-fiqh, Abd al-Aẓīm al-Dīb, Ed., First Edition, Vol. 1-2, Qatar University.

al-Juwaynī, Abd al-Malik (1981), Al-Shāmil fī uṣūl al-dīn, R. M. Frank, Ed., McGill University and Tehran University.

al-Juwaynī, Abd al-Malik (2016), Al-Irshād ilā qawāṭi al-adilla fī uṣūl al-itiqād, M. Y. Idrīs, Ed., Dār al-Nūr.

al-Māzarī, M. b. Alī (2000), Īḍāḥ al-maḥṣūl min Burhān al-uṣūl, Ammār al-Ṭālibī, Ed., Dār al-Gharb al-Islāmī.

al-Muqtaraḥ, Taqī al-Dīn (2014) Sharḥ Al-Irshād. Edited by Naziha Ma’arij. 2 vols. Tatouan: al-Rābiṭa al-Muḥammadiyya.

al-Rāzī, F. al-Dīn (2002), Manṭiq al-Mulakhkhaṣ, A. Qaramelki, Ed., Imam Sadeq.

al-Rāzī, F. al-Dīn (2015), Nihāyat al-uqūl, S. Foudeh, Ed., First Edition, Vol. 1-4, Dār al-Dhakhāʾir.

al-Sanūsī, M. b. Y., & al-Ḥāmidī, I. (2011), Sharḥ al-Sanusiyya al-Kubrā, Reprint of 1354/1935 Edition, Dār al-Imām al-Rāzī.

al-Shahrastānī, Abd al-Karīm (1934), Nihāyat al-iqdām fī ilm al-kalām, A. Guillaume, Ed., Oxford University Press.

al-Subkī, T. al-Dīn. (n.d.), Ṭabaqāt al-Shāfiiyya al-kubra, M. al-Ṭanāḥī & Abd al-Fattāḥ al-Ḥilū, Eds., Vol. 1-10, Dār Iḥyā al-Turāth al-Arabī.

al-Taftāzānī, S. al-Dīn (1860), Sharḥ al-Maqāṣid, Dār al-Ṭibāa al-Āmira.

al-Taftāzānī, S. al-Dīn (2012), Sharḥ al-Shamsiyya, J. Bassām, Ed., Dār al-Nūr.

al-Bāqillānī, Abū Bakr (1957), Al-Tamhīd, R. J. McCarthy, Ed., al-Maktaba al-Sharqiyya.

Bartha, P. (n.d.), Analogy and Analogical Reasoning, Ii E. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Spring 2019), [https://plato.stanford.edu/archives/spr2019/entries/reasoning-analogy/]

Benevich, F. (2016), The Classical Asharī Theory of Aḥwāl: Juwaynī and His Opponents, Journal of Islamic Studies, 27(2), p. 136-175.

El-Rouayheb, K. (2016), Review of The First Islamic Reviver by Kenneth Garden and Al-Ghazālī’s Moderation in Belief, trans. Aladdin Yaqub, Journal of the American Oriental Society, 136(2), p. 456-457.

Gelenbevī, I., & al-Dawānī, J. al-Dīn (1898), Ḥāshiya alā Sharḥ al-Aqāʾid al-aḍudiyya (Vol. 2), Der Saadet.

Ibn al-Tilimsānī, S. al-Dīn (2011), Sharḥ Maālim uṣūl al-dīn, N. Ḥammādī, Ed., Maktabat al-Maārif.

Ibn Fūrak, A. B. M. b. al-Ḥasan (1987), Mujarrad Maqālāt al-Asharī, D. Gimaret, Ed., Dār al-Mashriq.

Ibn Sina, Abū ꜥAlī al-Ḥusayn (1952), Al-Shifāʾ: Al-Qiyas, ed. Saeed Zayid, Vol. 4. 10 vols. Cairo, The General Egyptian Library.

Shihadeh, A. (2005), From al-Ghazali to al-Razi: 6th/12th Century Developments in Muslim Philosophical Theology, Arabic Sciences and Philosophy, 15, p. 141-179.

Shihadeh, A. (2013), The Argument from Ignorance and its Critics in Medieval Arabic Thought, Arabic Sciences and Philosophy, 23, p. 171-220.

Young, W. E. (2019a), Concomitance to Causation: Arguing Dawarān in the Proto-Ādāb al-Baḥth, in P. Adamson (Ed.), Philosophy and Jurisprudence in the Islamic World, De Gruyter, p. 205-281.

Young, W. E. (2019b), Islamic Legal Theoretical and Dialectical Approaches to Fallacies of Correlation and Causation (7th-8th/13th-14th centuries), in M. Bedir & R. Gleave (Eds.), Islamic Legal Theory: Intellectual History and Usul al-Fiqh. Forthcoming.

Haut de page

Notes

1 The Arabic expression in the recently edited text is mā yajibu itimāduhu alā al-qaṭ wa-l-yaqīn, which means ‘what is obligatory to rely upon with proof and certainty,’ but ‘reliability’ here is clearly ambiguous, and as is well known, ilm al-kalām has a normative aim of supporting beliefs derived from revelation through developing a philosophical system which includes epistemology, natural philosophy, and natural theology. The Arabic itimād is very close to the Arabic itiqād (which means ‘belief’), and it seems to be a case of misreading the manuscript or scribal error.

2 al-Baghdādī, Abd al-Qāhir (2019), Iyār al-naẓar fī ilm al-jadal, Kuwait, al-Asfār, Aḥmad Arrūbī (ed.) p. 141.

3Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 151.

4Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 155.

5Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 180.

6Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 282-284.

7Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 284.

8Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 284.

9  Bartha, Paul (2019), "Analogy and Analogical Reasoning", The Stanford Encyclopedia of Philosophy (Spring 2019 Edition), Edward N. Zalta (ed.), [https://plato.stanford.edu/archives/spr2019/entries/reasoning-analogy/]

10 This is not necessarily the case of course. It may be instead that one identifies an essential property from a single instance, and then, by virtue of the fact that it is essential, judges that all individuals of this type have the same property.

11 Ibn Fūrak, A. B. M. b. al-Ḥasan (1987), Mujarrad maqālāt Abī l-ḥāsan al-Asharī, Beirut, Dār al-Mashriq, Gimaret (ed.), p. 289, lines 4-11; cf. Bāqillānī, Abū Bakr (1957) Kitāb al-Tamhīd, Beirut, al-Maktaba al-Sharqiyya, McCarthy (ed.), p. 12-13.

12 Which for Asharī is at root a Divine Convention, cf. Mujarrad, p. 41-42.

13Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 284.

14Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 297.

15Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 297-8.

16Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 298.

17 Tāj al-Dīn al-Subkī (n.d.) Ṭabaqāt al-Shāfiiyya al-kubrā, Cairo, Dār Iyāʾ al-Turāth al-Arabī, Muḥammad al-Ṭanāḥī and Abd al-Fattāḥ al-Ḥilū (eds.), vol.v, p. 192.

18 Abū l-Maālī al-Juwaynī (1978), al-Burhān fī uṣūl al-fiqh, Qatar, Qatar University, Abd al-Aẓīm al-Dīb (ed.), p. 126-127.

19 Baghdādī

distinguishes two types of ‘collection and division’: exhaustive and non-exhaustive. Exhaustive divisions are ones where the options available are logically exhaustive of all possibilities. Therefore, eliminating all but one yields certainty in the remaining disjunct. Non-exhaustive divisions, on the other hand, cannot yield the same kind of certainty, because the initial set of divisions is discovered through induction. Therefore, it always remains a logical possibility of a further option. This kind of inference is also deployed in combination with analogical arguments, but I will not discuss it in detail here in consideration of space, but I will make reference to it below in the discussion on Ibn Sīnā’s critique of the analogical inference.

20Abd al-Malik al-Juwaynī (1978), Burhān, 126; cf. Abd al-Qāhir al-Baghdādī, (2019), Iyār, p. 284.

21Abd al-Malik al-Juwaynī (1981), al-Shāmil fī uṣūl al-dīn, Tehran, Tehran University, R. M. Frank (ed.), p. 60-69; cf. Juwaynī (2016), al-Irshād ilā qawāṭi al-adilla fī uṣūl al-itiqād, Amman, Dār al-Nūr, M. Yūsuf Idrīs (ed.) p. 129-133.

22Abd al-Malik al-Juwaynī (1978), Burhān, p. 127.

23Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 350.

24Abd al-Malik Juwaynī (1978), Burhān, p. 127; cf. Abd al-Malik Juwaynī (1981), al-Shāmil, p. 60-61. Note: the term muaṭṭila does not normally denote ‘atheist’ per say, but merely anyone who ‘denies’ something. The context in the citation above, however, indicates that the term refers to a group of thinkers who denied the existence of God.

25Abd al-Malik al-Juwaynī (1981), al-Shāmil, p. 61.

26Abd al-Malik al-Juwaynī (1981), al-Shāmil, p. 61.

27Abd al-Malik al-Juwaynī (1981), Shāmil, p. 62.

28Abd al-Malik al-Juwaynī (1981), Shāmil, p. 62. It would appear that this thought experiment forms the basis of the famous Ḥayy ibn Yaqẓān.

29Abd al-Malik Juwaynī (1978), Burhān, p. 127-128; cf. Abd al-Malik Juwaynī (1981), Shāmil, p. 63-65; Abd al-Malik Juwaynī (2016), Irshād, p. 129-130.

30 A. B. M. b. al-Ḥasan Ibn Fūrak (1987), Mujarrad, p. 80.23-25.

31 The term dalāla above is not identified as common elements. Instead, each one of those 4 identified common elements has its own signification (dalāla) which makes it fit to play the role of common element in an analogical argument. It may seem strange, after all, to consider ‘evidence’ as a common element in the same way that the other common elements apply, unless we consider ‘evidence’ as an external concomitant of the source/target entities.

32Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 129-131.

33Abd al-Malik al-Juwaynī (1981), Shāmil, p. 63.

34Abd al-Malik al-Juwaynī (1978), Burhān, p. 128.

35Abd al-Malik al-Juwaynī (1981), Shāmil, p. 63.

36 This is a previous book from within the Shāmil, which has been partially edited in the edition by Nashshār, see Abd al-Malik al-Juwaynī (1969), al-Shāmil fī uṣūl al-dīn, Alexandria, Manshaʾat al-Maārif, Alī Sāmī al-Nashshār (ed.), p. 629-716.

37 By the term biconditional, I mean the necessary co-presence and co-absence of the two elements.

38Abd al-Malik al-Juwaynī (1981), Shāmil, p. 63. The term ḥukm is often translated as ‘judgement’ or ‘ruling’, but in these contexts, I think that the term ‘property’ or even ‘predicate’ is more suitable. The reason is this: the illa or ‘cause’ is a real property that causes the ḥukm. In other words, there is a real metaphysical connection between illa and ḥukm. The term ‘judgement’, meanwhile, is often used to refer to a linguistic act or a belief or otherwise that exists in an observer. But certainly the illa here does not necessitate a belief or a judgement in observers, but rather, necessitates a further property or predicate in the entity which is the subject of that cause.

39Abd al-Malik al-Juwaynī (1969), Shāmil, p. 631.

40 For more on the aḥwāl, see Fedor Benevich (2016), “The Classical Asharī Theory of Aḥwāl: Juwaynī and His Opponents,” Oxford, Oxford University Press, p. 136-175; and Benevich (2018), “The Metaphysics of Muḥmmad b. Abd al-Karīm al-Shahrastānī (d.1153): Aḥwāl and Universals,” Bonn, Bonn University Press, Abdelkader Al Ghouz (ed.), p. 327-355.

41Abd al-Malik al-Juwaynī (1969) Shāmil, p. 646.

42Abd al-Malik al-Juwaynī (1969), Shāmil, p. 644-645. Of course, this inference could be denied by the ḥāl skeptic, but since the main opponent on the question of the reality of God’s attributes are the Baṣran Mutazila who accepted the existence of abstract modes yet denied the necessity of grounding them in an underlying property, the inference was effective.

43Abd al-Malik al-Juwaynī (1981), Shāmil, p. 64.

44Abd al-Malik al-Juwaynī (1978), Burhān, p. 128; Abd al-Malik al-Juwaynī (1981), Shāmil, p. 64.

45Abd al-Malik al-Juwaynī (1981), Shāmil, p. 63; cf. Burhān, p. 128.

46Abd al-Malik al-Juwaynī (1978), Burhān, p. 128.

47Abd al-Malik al-Juwaynī (1981), Shāmil, p. 64.

48Abd al-Malik al-Juwaynī (1981), Shāmil, p. 64. That is, unlike the first two cases which are biconditional, and the third case where one may only infer from conditioned property to condition, in the case of signifier/evidence, one can only infer from signifier to signified, but not from the absence of the signifier to the absence of the signified.

49Abd al-Malik al-Juwaynī (1978), Burhān, p. 128.

50Abd al-Malik al-Juwaynī (1981), Shāmil, p. 65.

51Abd al-Malik al-Juwaynī (1978), Burhān, p. 129-131.

52Abd al-Malik al-Juwaynī (1978), Burhān, p. 129-130.

53Abd al-Malik al-Juwaynī (1981), Shāmil, p. 28.

54Abd al-Malik al-Juwaynī (1978), Burhān, p. 130.

55 For example: Abū l-Qāsim al-Anṣārī (2010), al-Ghunya fī ilm al-kalām, Cairo, Dār al-Salām, Muṣtafā Abd al-Hādī (ed.), vol. 1, p. 519-520; Abd al-Karīm al-Shahrastānī (1934), Nihāyat al-iqdām fī ilm al-kalām, London, Oxford University Press, Guillaume (ed.), p. 182-183; Ibn al-Tilimsānī (2011), Sharḥ Maālim uṣūl al-dīn, Beirut, Maktabat al-Maārif, Nizār Ḥammādī (ed.), p. 282-284; Sad al-Dīn al-Taftāzānī (1860), Sharḥ al-Maqāṣid, Istanbul, Dār al-Ṭibāa al-Āmira, vol. 2, p. 54-55; Muḥammad b. Yūsuf al-Sanūsī (1935), Sharḥ al-Sanūsiyya al-kubrā, Cairo, p. 220-222; Ismāīl Gelenbevī (1898), Ḥāshiya alā Sharḥ al-Aqāʾid al-aḍudiyya, Istanbul, Der Saadet, vol. 1, p. 272-273.

56 Cf. Abd al-Malik al-Juwaynī (1969), Shāmil, p. 660-665.

57 Ibn Sīnā (1964), al-Shifā, Kitāb al-Qiyās, Cairo, al-Maktaba al-āmma, Saīd Zāyid (ed.), vol. 4, p. 568-572.

58 Ibn Sīnā (1964), al-Shifā, Kitāb al-Qiyās, vol. 5, p. 573-580

59 Abū Naṣr al-Farābī (2014), al-Manṭiq inda al-Farābī: Kitāb al-Qiyās, Beirut, Dār al-Shurūq, Rafīq Ajam (ed.), vol. 2 p. 47-48.

60 Abū Naṣr Farābī (2014), Kitāb al-Qiyās, p. 48.

61 Cf. Ghazālī (2016), Miyār al-ilm, Beirut, Dār al-Minhāj, p. 190-196; and Fakhr al-Dīn al-Rāzī (2002), Manṭiq al-Mulakhkhaṣ, Tehran, Imam Sadeq, Qaramelki (ed.), p. 339-340.

62 Cf. Schmitt, Jens Ole, (in this volume), “Preferring Formal Language over the Face? Avicenna on the Physiognomical Syllogism. Some Observations.”

63 Ibn Sīnā (1964), Qiyās 9.24, p. 574.12-13.

64 Ibn Sīnā (1964), Qiyās, 9.24, p. 575.14-576.4.

65Abd al-Qāhir al-Baghdādī (2019), Iyār, p. 286.

66Abd al-Malik al-Juwaynī (1981), Shāmil, p. 66-67.

67Abd al-Malik al-Juwaynī (1981), Shāmil, p. 66-67.

68Abd al-Malik al-Juwaynī (1978), Burhān, p. 131.

69Abd al-Malik al-Juwaynī (1978), Burhān, p. 131.

70Abd al-Malik al-Juwaynī (1978), Burhān, p. 157-158.

71 Ibn Sīna (1964), Qiyās p. 9.24, p. 576-577.

72 Ibn Sīnā (1964), Qiyās p. 9.24, p. 576-577.

73 F. al-Dīn Rāzī (2002), Mulakhkhaṣ, p. 339-340.

74Abd al-Malik al-Juwaynī (1969), Shāmil, p. 661. The Arabic in the Nashshār edition reads: idhā lam yakun mujarrad al-iṭṭirād wa-l-inikās ālimān fī intiṣāb al-shayʾ illatan… with the term ‘ālim’ here literally meaning knower or world depending on how one reads it. Consulting the Tehran Manuscript however, it reads alām or alā; if the first, it means sign, which is more appropriate to the context, and if it is the second, then it would appear that this is a scribal error with the complete word being alāma, which correctly recurs frequently in the chapter, e.g., p. 662 of the Nashshār edition.

75 Cf. Young, Walter E. (2019), “Concomitance to Causation: Arguing Dawarān in the Proto-Ādāb al-baḥth,” in Peter Adamson, ed., Philosophy and Jurisprudence in the Islamic World, p. 205-271. Berlin, Boston: De Gruyter; and Young, Walter E. (2019), “Islamic Legal Theoretical and Dialectical Approaches to Fallacies of Correlation and Causation (7th-8th/13th-14th Centuries).” [Edited volume from the conference “Islamic Legal Theory: Intellectual History and Uṣūl al-Fiqh,” Istanbul, 2019]. Eds. R. Gleave (Exeter) and M. Bedir (Istanbul). [Forthcoming].

76Abd al-Malik al-Juwaynī (1969), Shāmil, p. 661.

77Abd al-Malik al-Juwaynī (1969), Shāmil, p. 661.

78Abd al-Malik al-Juwaynī (1969), Shāmil, p. 661-663.

79Abd al-Malik al-Juwaynī (1969), Shāmil, p. 664-665.

80 A. l-Qāsim Anṣārī (2010), al-Ghunyā, p. 520.

81 A. l-Qāsim Anṣārī (2010), al-Ghunyā, p. 521. Barring teleportation and the interpenetration of like bodies.

82 A. l-Qāsim Anṣārī (2010), al-Ghunyā, p. 523-529.

83 A. l-Qāsim Anṣārī (2010), Ghunyā, p. 519-520.

84Abd al-Karīm Shahrastānī (1932) Nihāya, p. 182.

85 A. Ḥāmid Al-Ghazālī (2016), Miyār al-ilm, p. 190.

86 A. Ḥāmid Ghazālī (2016), Miyār, p. 190.

87 A. Ḥāmid Ghazālī (2016), Miyār, p. 193.

88 A. Ḥāmid Ghazālī (2016), Miyār, p. 193.

89 A. B. M. b. al-Ḥasan Ibn Fūrak (1987), Mujarrad, p. 187.

90 A. Ḥāmid Ghazālī (2016), Miyār, p. 194-196.

91Abd al-Karīm Shahrastānī (1932), Nihāya, p. 182-183.

92Abd al-Karīm Shahrastānī (1932), Nihāya, p. 186.

93 M. b. Alī Al-Māzarī (2000), Īḍāḥ al-maḥṣūl min Burhān al-uṣūl, Tunis, Dār al-Gharb al-Islāmī, Ammār al-Ṭālibī (ed.), p. 106.

94 Taqī al-Dīn al-Muqtaraḥ (2014), Sharḥ al-Irshād, Tatouan, al-Rābia al-Muammadiyya, vol. 1, p. 314-315; cf. al-Muqtaraḥ, Ḥāshiya alā al-Burhān, Iraqi Museum, ms. 996, fol. 41.

95 Taqī al-Dīn al-Muqtaraḥ (2014), Sharḥ al-Irshād, vol. 1, p. 314.

96 Taqī al-Dīn al-Muqtaraḥ (2014), Sharḥ al-Irshād, vol. 1, p. 286, p. 326.

97 Taqī al-Dīn al-Muqtaraḥ (2010), Sharḥ al-Burhāniyya, Beirut, Maktabat al-Maārif, Nizār Ḥammādī (ed.), p. 67-69.

98 Al-Abyārī (2013), al-Taḥqīq wa l-bayān fī sharḥ al-Burhān, Kuwait, Dār al-iyāʾ, vol. 1, p. 433.

99 Abyārī (2013), Sharḥ al-Burhān, vol. 1, p. 433.

100 A. Shihadeh (2005), “From Al-Ghazālī to Al-Rāzī,” in Arabic Sciences and Philosophy, p. 165-168.

101 K. El-Rouayheb (2016), Review of The First Islamic Reviver by Kenneth Garden and Al-Ghazālī’s Moderation in Belief, trans. Aladdin Yaqub, Journal of the American Oriental Society, 136(2), p. 456-457.

102 A. Shihadeh (2005), “From Al-Ghazālī to Al-Rāzī”, Arabic Sciences and Philosophy, p. 165-168.

103 A. B. Bāqillānī (1998), al-Taqrīb wa l-Irshād, Beirut, Al-Resalah Publishers, Abd al-amīd Abū Zayd (ed.), vol. 1, p. 228-231.

104 A. Shihadeh (2013), “The Argument from Ignorance and Its Critics,” Arabic Sciences and Philosophy, p. 191-195.

105 F. al-Dīn Rāzī (2015), Nihāyat al-uqūl, Beirut, Dār al-Dhakhāʾir, Saīd Fūdā (ed.), vol. 1, p. 136.

106 F. al-Dīn Rāzī (2015), Nihāya, vol. 1, p. 137.

107 F. al-Dīn Rāzī (2015), Nihāya, vol. 2, p. 235.

108 Ṣadr al-Sharīa al-Maḥbūbī, Sharḥ Tadīl al-ulūm, Topkapı Sarayı Müzesi Emanet Hazinesi, ms. 1669, fol. 108; cf. Avicenna in Young, “Concomitance to Causation,” 260 ff.

109 Ṣadr al-Sharīa, Sharḥ Tadīl al-ulūm, fol. 108.

110 Sad al-Dīn al-Taftāzānī (2012), Sharḥ al-Shamsiyya, ed. Jādullah Bassām, Amman, Dār al-Nūr, p. 370-372.

111 Cf. Sad al-Dīn al-Taftāzānī, in Young, “Concomitance to Causation,” p. 265.

112 Sad al-Dīn al-Taftāzānī (2012), Sharḥ al-Shamsiyya, p. 372.

113 Sad al-Dīn al-Taftāzānī (2012), Sharḥ al-Shamsiyya, p. 372.

114 Sad al-Dīn al-Taftāzānī (1860), Sharḥ al-Maqāṣid, vol. 2, p. 54-55.

115 Jalāl al-Dīn al-Dawānī (1898), Sharḥ al-Aqāʾid al-Aḍudiyya, vol. 1, p. 273.

116 I. Gelenbevī (1898), Sharḥ al-Aqāʾid al-Aḍudiyya, vol. 1, p. 272-273.

117 I. Gelenbevī (1898), Sharḥ al-Aqāʾid al-Aḍudiyya, vol. 1, p. 273; cf. Taftāzānī (1860), Sharḥ al-Maqāṣid, vol. 2, p. 54-55.

118 I. Gelenbevī (1898), Sharḥ al-Aqāʾid al-Aḍudiyya, vol. 1, p. 273.

119 A. b. Y. al-Harawī (2019), Sharḥ Tahdhīb al-manṭiq, Amman, Dār al-Nūr, Abd al-Ḥamīd al-Turkmānī (ed.), p. 206-207. Interestingly, the editor Turkmānī failed to understand the point made by al-Harawī, and merely repeats the false view that ‘juridical analogies’ do not yield certainty, despite the explicit statement by the author in the text itself.

Haut de page

Pour citer cet article

Référence électronique

Abdurrahman Ali Mihirig, « Analogical Arguments in the Kalām Tradition: Abū l-Maālī al-Juwaynī and Beyond  »Methodos [En ligne], 22 | 2022, mis en ligne le 01 mai 2022, consulté le 29 mars 2024. URL : http://journals.openedition.org/methodos/9004 ; DOI : https://doi.org/10.4000/methodos.9004

Haut de page

Auteur

Abdurrahman Ali Mihirig

Ludwig-Maximilian University

Haut de page

Droits d’auteur

CC-BY-NC-ND-4.0

Le texte seul est utilisable sous licence CC BY-NC-ND 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

Haut de page
Search OpenEdition Search

You will be redirected to OpenEdition Search