Our sincere appreciation goes to Leone Gazziero (CNRS, UMR 8163 « Savoirs, Textes, Langage », Université de Lille), chief editor of Methodos, and to members of the editorial board who enthusiastically supported the present special issue. Thanks are also due to Laurent Cesalli (Université de Genève, Département de Philosophie) and Tony Street (University of Cambridge, Faculty of Divinity), and once again Leone Gazziero, who, in their capacity as leaders of the ERC-Generator project “Logic in Reverse: Fallacies in the Latin and the Islamic Traditions,” fostered the research leading to this present study. We would also like to thank Prashant Keshavmurthy (McGill University, Institute of Islamic Studies) and Asad Q. Ahmed (University of California, Berkeley, Dept. of Middle Eastern Languages and Cultures), the first for positing a stimulating question on abduction in the Islamicate logical / dialectical traditions, and the second for valuable comments in the subsequent interchange. Finally, our heartfelt thanks go to Saïs Mandjee, for translating our abstract into French.

Introduction

2The aim of qiyās is to provide a rational ground for the application of a juridical ruling to a given case which has not been directly pronounced upon in the primary juridical sources (i.e., the Qurʾān, the Sunna [Prophet’s example], and Ijmāʿ [consensus]). It combines heuristic (and/or hermeneutic) moves with logical inferences; and archetypal qiyās—that is, qiyās al-ʿilla, or “correlational inference of the occasioning factor / cause”—adheres to the following pattern:

In order to establish whether or not a given juridical ruling (ḥukm) applies to a novel or contended case, called the branch-case (farʿ),

1. We look for a relevant, authoritatively determined root-case (aṣl) bearing that ruling in the primary sources of law (Qurʾān, Sunna, and Ijmāʿ).

2. • 3 In referring to the ʿilla as the “occasioning factor” we follow Weiss, Bernard (2010), The Search f (...)

   We next attempt to determine the property (waṣf) or set of properties in the root-case which constitutes the ʿilla—that is, the occasioning factor (or legal cause, or ratio legis)3 which gives rise to its ruling.

3. If we may determine that this property occasions the ruling, and that it is shared by the branch-case, we may infer that it is equally productive of that ruling in the branch-case.

4. The novel or contended branch-case thus falls under that juridical ruling, and the range of its application is extended.

When the ʿilla (occasioning factor) is made explicit by the sources, or is capable of being rationally inferred by adequately identifying the relevant property or set of properties, we may proceed with a “correlational inference of the occasioning factor” (qiyās al-ʿilla).

3A classic example is to reason that date liquor, being intoxicating just like grape wine, is therefore also prohibited like grape wine. As identified by canonical analysis, the four elements in this argument are:

• the farʿ : the branch-case under consideration (date liquor);

• the aṣl : the root-case verified by the primary sources (grape wine);

• the ʿilla : the occasioning factor they have in common (intoxication); and

• the ḥukm : the legal qualification (prohibition) which is therefore also common, inferred in the case of date liquor via the source-verified case of grape wine.

From this it is evident that the key procedure underpinning this form of qiyās is identifying the ʿilla: the occasioning factor giving rise to prohibition in our example aṣl of grape wine. From a different perspective, applying the general schema intoxicating drinks should be forbidden to the farʿ of date liquor occasions its interdiction.

4When the occasioning factor (ʿilla) is neither made explicit by the sources nor capable of being rationally inferred, however, we might next resort to “correlational inference of indication” (qiyās al-dalāla). Here, in lieu of the ʿilla, one pinpoints relevant parallelisms between other rulings known to be shared by aṣl and farʿ, thus inferring that whatever the ʿilla may be, it must be shared in the case at hand (as it must also have been shared in the other cases). Should even this prove infeasible, we might finally resort to the (highly contentious) “correlational inference of resemblance” (qiyās al-shabah), which is based merely on the presence of properties shared but either of indeterminable or nonexistent causative efficacy.

5Thus, qiyās al-dalāla and qiyās al-shabah—both of which, far more than qiyās al-ʿilla, merit the label “argument by analogy” (or, better yet, “argument a pari”)—are put into action when the ʿilla grounding the application of a given ruling is not known. The plausibility of a conclusion attained by parallelism between other shared rulings (qiyās al-dalāla) is considered to be of a higher epistemic degree than a conclusion obtained by mere resemblance in respect to some set of (relevant) properties (qiyās al-shabah).Conclusions by either have a lower epistemic standing than conclusions inferred via a known, pinpointed, and shared occasioning factor (qiyās al-ʿilla).

6A cardinal feature of al-Shīrāzī’s take on qiyās al-ʿilla is his particular notion of efficiency (taʾthīr) which tests whether the property P purported to be efficient in occasioning the juridical ruling at stake is indeed so. For al-Shīrāzī, taʾthīr consists of two complementary procedures:

• co-presence (ṭard): whenever the property is present, the ruling is also present, and

• co-absence (ʿaks): whenever the property is absent, the ruling is also absent.

10In brief, a deeper logical and linguistic analysis of the different existential modes of dawarān strongly suggests that we should distinguish property (or phenomenon) and ruling (or effect) as types (concepts or propositions linguistically expressed by a sentence) as opposed to tokens (instantiations of the type; the real, ontological events that verify the proposition). Rather than simply reading “in existence” as “true” and “in nonexistence” as “false” within the framework of a classical truth-functional formal semantics, we here read dawarān as shaped by a finer-grained structure—one in which a dawarān test confirms:

1. concomitance “in existence” if any token of the property (or phenomenon) type triggers a token of the corresponding ruling (or effect) type, and

2. concomitance “in nonexistence” if any absence of a token of the property (or phenomenon) type triggers an absence of a token of the corresponding ruling (or effect) type.

This analysis complements the recent work of Shahid Rahman and Muhammad Iqbal on qiyās (2019), wherein they applied contemporary Type-Theoretical grammar as developed by Aarne Ranta (1994), based on Per Martin-Löf’s (1984) Constructive Type Theory and its inferential take on the principle of propositions as sets of types.

11The CTT-approach allows us not only to distinguish the types of property (or phenomenon) and ruling (or effect) from their tokens or instances, but also to identify the efficient occasioning process as a function which takes some particular token of the ʿilla (e.g., my consumption of an intoxicating drink today), and renders a token (the interdiction of this particular action of my consumption of an intoxicating drink today) of the general ruling type (consuming intoxicating drinks is forbidden). As a result, we may interpret the Muslim legal theoreticians’ ʿilla as both:

1. the material cause / occasioning factor / ratio legis, which refers to the property (or compound of properties) P as a type, and

2. • 9 This function ʿilla(x): H(x) (x: P) may be read as follows: “Given the property or properties P, th (...)
   • 10 In fact, in an Aristotelian framework we might identify a case’s property (or properties) as the ma (...)

   the efficient causal / occasioning process: the function ʿilla(x): H(x) (x: P)9 that links the existence of the property P with the existence of the ruling H – i.e., the function that relates tokens x of the property with tokens of (applications of) the ruling.10

12This analysis also allows us to elucidate the epistemological concept of causation at work in dawarān, which—despite his denial of the utility of concomitance alone in determining the ʿilla in contexts of juristic qiyās—appears to be paralleled by al-Ghazālī’s approach to natural causality, in contrast to Ibn Sīnā’s essentialist approach. It is the non-essentialist approach to causation that requires a singling out of the procedure that links the efficient property to the occasioned effect. And since, in al-Ghazālī’s epistemology, causation is explained by neither essential nor accidental active and passive powers, the primary focus of the causal justification process (taʿlīl) remains the link between presences of the property and presences of the effect—that is, our second interpretation of the ʿilla as efficient process and function.

14In conclusion, we will suggest that the formulation “in existence and in nonexistence,” in the context of dawarān, expresses the very essence of the prescriptive character of legal norms—in general, not only within the Islamicate tradition; and it does so, for the most part, by way of the following features:

1. legal norms’ comprising an intertwining of type and token, whereby the type supplies the prescription’s generality and hypothetical character, while the token constitutes the matter to which the prescription is applied—that is, the realization of the type (note the hypothetical character of the prescription does not require the involved token’s actualization or presence);

2. juridical rulings’ dependence upon causal / occasioning properties, whereby the efficient sense of ʿilla not only expresses the link relating a property (or compound of properties) to its specific (legal) effect, but constitutes the primary objective of the modes of causal justification (masālik al-taʿlīl);

3. • 12 In fact, it is the distinction between the property and the efficient process of relating presences (...)

   the equal applicability of (2) to natural causality, whereby a type of effect is said to depend upon a type of causal property in such a way that some physical and/or chemical process realizes a token of the effect;12

4. dawarān’s being, in the first place, a test—that is, a verification procedure, rather than a conjunction of verified propositions; and

5. the conjectural epistemic status of the ʿilla being won through dialectical inquiry, and postulated—with special regard to the arguments brought forward in support—as the best explanation for the (legal) effect under consideration.

The ʿIlla as an Occasioning Factor with neither an Active nor Passive Nature

Towards an Account of Causation at Work in Dawarān

In Avicenna, like in Aristotle, the source of our knowledge of the essential active and passive powers of things is not nature and its observation but the separate active intellect. Sensual perception, Avicenna teaches, cannot lead to necessary judgments. It is important to note that induction only works if the active and passive powers that lead to causal connections are part of the essences of the things.

• 18 Notice that induction (istiqrāʾ) is distinguished from experimentation (tajriba); whereas istiqrāʾ (...)

When the active and passive powers that necessitate the causal connection are not part of the essences of the things, Avicenna mandates the use of experimentation (tajriba). An example that Avicenna and al-Ghazālī both mention is that in medicine, we witness that scammony causes purgation in the gallbladder. According to Avicenna, the relationship between scammony and the purgation of bile is not due to an active power that is part of the essence of scammony. Rather, the effect is due to an “inseparable accident” (ʿaraḍ lāzim) or a proprium (khāṣṣa) of scammony, meaning an accident that inheres permanently and is therefore an inseparable part of it. Since the cause of this laxative effect is an accidental characteristic, we cannot know it through induction (istiqrāʾ).18 In this case, experimentation (tajriba) leads us to conclude that the accident of causing this laxative effect inheres in scammony. The repeated observation of this connection establishes that there is something either in scammony’s nature or just “with it” (maʿahu) that causes—at least in our lands, Avicenna adds—purgation of bile.

[….]

In Avicenna’s view, experimentation informs us that scammony has a purging effect, yet it does not allow us to conclude how this effect occurs. Unlike induction, it does not provide the underlying causal explanation. Experience thus does not provide scientific knowledge (Greek episteme, Arabic ʿilm) in the strict Aristotelian sense of it being both necessary and explanatory.

[….]

• 19 NB: this corresponds to the legal theorists’ notion of the māniʿ, or impediment, the presence of wh (...)

This method often forces the scientist to limit his or her results to the conditions he or she observed, such as when Avicenna says that scammony has the observed effect “in our lands.” Limitations, such as the acknowledgment that scammony may not have its purging effect in other climates, are very important in Avicenna’s theory of experience.19 They are a result of the fact that we are only dealing with a cause that is an accident in scammony, and not a part of its essence.

19More generally, integrating into the method of law-making tools that prompt the withdrawal of causation claims by justifying exceptions or accommodating special conditions shapes a body of law’s dynamic nature and appears to counter any assumption of essential (and therefore static) causative powers. Ultimately, it is the body of law and the dialectical endeavour expanding and refining it that ground causation claims within the limits of a resource-bounded human knowledge. Thus, if a general framework comprising causes / occasioning factors for all of legal, theological, and natural contexts is to be developed, then appealing to active and passive powers may be rather more hindering than helpful. And if properties alone are not sufficient to explain causation, then determining causal efficiency must remain our focus; more precisely: justifying claimed links between occasioning properties and occasioned effects.

21The upshot of this kind of analysis is that it allows us to make explicit the distinction between (1) the property constituting the material-causal occasioning factor, and (2) the actual, efficient-causal occasioning factor. In other words, according to this view, it is useful to distinguish between the property (waṣf, pl. awṣāf) and the properly efficient sense of ʿilla; and the procedure is one that transforms instances of occasioning factors into applications of the effect to these instances.

22Here, we have slightly adapted a schema which is more fully introduced in Rahman & Young (Forthcoming, 2022), in order to elucidate further the analyses developed in Rahman and Iqbal (2018, p. 80-84); Rahman, Iqbal, & Soufi (2019, p. 31-40); and Iqbal (2021). And since we would like to underscore apparent efforts to unify legal and natural causation, we will employ the technical components of later, logicized dawarān theory, incorporating the term madār for the concomitant presumed cause and the term dāʾir for the concomitant presumed effect.

The formal expression above may be read as follows:

Given some concomitant factor presumed to be capable of material causality / occasioning power (the madār), i.e., a concomitant, suitable property, there is some efficient cause / occasioning factor, ʿilla(x), that occasions the concomitant presumed effect (the dāʾir), i.e., the concomitant juridical ruling.

Thus, ʿilla(x) is a procedure (a function) that relates presences of what is presumed to be the concomitant material cause / occasioning factor with presences of what is presumed to be its concomitant effect.

24Notice that the expression is a hypothetical judgement, not a hypothetical proposition (that is, it is neither an implication nor a universal proposition). Neither the presence of the antecedent (x: madār) nor the presence of the consequent dāʾir(x) is presumed: it is purely hypothetical. As explained below, we can take a further step and express the hypothetical as the universal; i.e., for any presence of what is presumed to be the concomitant material cause / occasioning factor, there is a method that relates those presences with presences of what is presumed to be its concomitant effect. This amounts to raising the hypothetical to a universal law. The method / efficient occasioning factor verifies the universal law; i.e., if such a method can be exhibited, then the universal is said to be verified. For example: if the ruling Every consumption of an intoxicating drink is forbidden is verified, then there must be some legal procedure, enacted by some legal authority, that leads to the interdiction of any performance of such a consumption.

Dawarān as a Procedure

25We can now apply this same kind of analysis to the full notion of dawarān. Turning our attention first to co-presence (ṭard), or concomitance “in existence,” we take as a given that ṭard is whenever the madār is present, the dāʾir is also present. Thus, for any m that actualizes the madār, ṭard verifies that m produces the dāʾir.

26Notice that this analysis makes explicit the fact that ṭard is a verification procedure which takes instances of the presumed cause and produces instances of its presumed effect. If the procedure triggers the effect for any instance of the madār, then we can assert that ṭard co-presence has been verified. In fact, this verification procedure has a universal force: co-presence is verified, if the procedure confirms that all presences of the madār yield presences of the dāʾir:

The expression “λx.ṭard(x)” indicates that the procedure ṭard confirms that for each madār-presence a dāʾir-presence can be found. Thus, if λx.ṭard(x) can be shown to obtain, then the proposition that all presences of the madār yield presences of the dāʾir—i.e., (∀x: madār) dāʾir(x)—will be verified.

27Turning our attention now to co-absence (ʿaks), or concomitance “in nonexistence,” we see that it can be analysed in the same way. In brief, we take as a given that ʿaks is whenever the property is absent, the ruling is also absent. Thus, we obtain:

λy.ʿaks(y): (∀y: ~madār) ~dāʾir(y).

Here, the expression “λy.ʿaks(y)” indicates that the procedure ʿaks confirms that for each madār-absence a dāʾir-absence can be found. Thus, if λy.ʿaks(y) can be shown to obtain, then the proposition that all absences of the madār yield absences of the dāʾir—i.e., (∀y: ~madār) ~dāʾir(y)—will be verified.

28Since dawarān as a verification test consists of both the test of ṭard and the test of ʿaks, we may now pull all this together and express dawarān as the following conjunction:

dawarān: (∀x: madār) dāʾir(x) ∧ (∀y: ~madār) ~dāʾir(y)

or, as more explicitly comprising both methods of ṭard and ʿaks:

29In the preceding, the reader may have noticed the use of two quantifiers (x and y). This avoids the problem of asserting of the same x that it both has and does not have the occasioning property at the same time. One might object, however, that any dawarān test should concern the same case x both as having and as not having the considered property—albeit under different circumstances. For the aims of the present paper, we have left this to the dynamics of the dialogical structure, so that each choice of the interlocutor triggers a different dialogue (play), whereby the same case is examined under different contexts. By way of illustration, we might imagine an example in which one play concerns pressed juice with the quality of wine while the other concerns the same pressed juice with the quality of vinegar. It is in fact possible to encode this into a logical analysis, but a bit more complexity is required. There are several ways to accomplish this.

30One way is to assume that every x either enjoys or does not enjoy the occasioning property P; and if some y is identical to those instances of x that enjoy the property P, then the ruling H follows; and if some z is identical to those x that do not enjoy the property P, then some other ruling—here we will assume the opposite ruling ~H—follows. Let us consider the paradigmatic example-case of wine. We will assume that

1. every pressed juice x enjoys or does not enjoy the occasioning property P: being an intoxicating drink;

2. if any pressed juice y is identical to those pressed juices x that enjoy the property P, then it is forbidden for consumption;

3. and if any pressed juice z is identical to those pressed juices x that do not enjoy the property P, then it is allowed for consumption.

A formal expression of such an analysis, in which “A” stands for “intoxicating drink,” yields

31(∀x: A ∨ ~A) [ [ (∀y: A) left^∨(y)=x ⊃ Forbidden-for-consumption(y)] ∧ [ (∀z: ~A) right^∨(z)=x ⊃ Allowed-for-consumption(z)] ]

32Returning to al-Ghazālī, a well-known feature of his take on natural causation is that the link between cause and effect is ultimately grounded on God’s will, although some level of necessary knowledge can be nevertheless achieved by humans. In principle, however, this knowledge is for the most part situated at an epistemic level of judgement, not at an ontological level. This, in the main, is consonant with al-Ghazālī’s critique of mere concomitance as a test for causation, as developed in his legal theory. Here we may provide a brief summary, drawn from his famed legal-theoretical manual the Mustaṣfā (see the appendix for full Arabic text and translation).

33In his Mustaṣfā, al-Ghazālī approaches the methods of determining the ʿilla in a relatively standard fashion. First is affirming the cause / occasioning factor (ithbāt al-ʿilla) by way of transmitted textual indicants (al-adilla al-naqliyya), then by way of juristic consensus (ijmāʿ), then by way of rational inference and methods of drawing indication (al-istinbāṭ wa ṭuruq al-istidlāl). This latter category includes analytical disjunction and exclusion (al-sabr wa-l-taqsīm) and suitability (munāsaba), but not (according to al-Ghazālī) concomitance. Rather, concomitance belongs to the category which follows; namely, “the invalid methods for affirming the root-case’s cause / occasioning factor” (al-masālik al-fāsida fī ithbāt ʿillat al-aṣl). The second of these is co-presence (ṭard) alone, and the third is both co-presence and co-absence (al-ṭard wa-l-ʿaks).

35Moving on to the next invalid method, ṭard and ʿaks together fair no better. As exemplified by wine’s peculiar odour, which is both co-present and co-absent with wine’s prohibition, al-ṭard wa-l-ʿaks is no indicator of causality, only of some kind of linkage. Al-Ghazālī argues that this is because (as he has just shown) co-presence alone proves nothing in terms of causation, and adding co-absence brings no further efficacy since co-absence is not a necessary condition for legal causes (ʿilal sharʿiyya). Moreover, a property might be fully concomitant with a ruling not because it is the ʿilla, but because it is one of the ʿilla’s entailments, or one of its parts, or one of its conditions. However, al-Ghazālī considers the combination of analytical disjunction and exclusion (al-sabr wa-l-taqsīm) with al-ṭard wa-l-ʿaks to constitute a valid proof (ḥujja). Indeed, a valid proof is attained by combining it with ṭard alone, without ʿaks. Though one might err by missing a property in one’s analysis which turns out to be the ʿilla, the burden of exposing missed properties lies on the claimant.

37Merging this summary with his take on natural causality, it is clearly the case for al-Ghazālī that mere concomitance (whether ṭard alone, or al-ṭard wa-l-ʿaks) or experience (tajriba)—even if it might pave the way towards discovering occasioning factors / causes—is insufficient. Pinpointing causation requires additional balance-tipping procedures (viz. al-sabr wa-l-taqsīm), be they epistemological, hermeneutical, inferential, or procedural. Based on the example of al-Ghazālī, then, we might say that a process for formulating a general schema explaining the kind of causation at work in a given case cannot be won by dawarān or experience alone, though these may render collections of like cases, and, perhaps, a certain regularity. But the task of formulating explicitly the precise content and form of links between the madār-presences and dāʾir-presences involved in a causation claim brings any such schema out into the open, exposing it to challenges and demands for justification.

Taʿlīl, the Larger Task of Justifying the Cause, and “Arguments to the Best Explanation”

41This wider range of causal justificatory activities—including, but not limited to, the masālik al-taʿlīl—may, at a future date, be classified under certain groupings which coincide with the aims of the present study. Now we would only suggest a handful of categories (tentatively, and with no claim of being either collectively exhaustive or mutually exclusive); these would include:

• procedures and methods for choosing between alternatives and eliminating choices (e.g., al-sabr wa-l-taqsīm and al-istidlāl bi-l-taqsīm),

• for raising alternative explanations (e.g., muʿāraḍa), and

• for introducing exceptions and withdrawing conclusions (e.g., istiḥsān).

42Importantly, post-classical definitions and conceptions of dawarān popularized by such as Burhān al-Dīn al-Nasafī (d. 1288 CE) and Shams al-Dīn al-Samarqandī “completed” dawarān, as Young (2019, p. 281, n. 237) puts it:

by folding suitability (munāsaba) into it – as the ultimate expression of “rightness of causation” (ṣulūḥ al-ʿilliyya) – or [by considering] a refined version of dawarān alongside munāsaba and analytical division (al-sabr wa-l-taqsīm), as a streamlined trio of complementary and mutually supportive methods.

Together with what has just been argued, it is evident that the larger project of justifying one’s choice of occasioning factor as the most correct (or just, or reasonable) not only involved a whole gamut of methods and procedures, but also encouraged—or even required—their application in concert with one another.

43All told, then, we might posit that the outcome of such methods and their concatenation were, or should have been, productive of arguments that:

1. led to a grasping of the universality of Law in the particular;

2. had an explanatory power beyond statistical confirmation, providing the answer to how the occasioning factor fulfils the duty of causation;

3. outperformed competing solutions, in relation to the available epistemology, textual sources, and hermeneutics; and

4. were sensitive to dynamic information inputs.

Points 1 and 2 relate to the above-mentioned notion that experience (tajriba) alone cannot yield universal laws and, as we will discuss further below, that istiqrāʾ should not be reduced to a merely statistical induction. This is crucial if applications of dawarān are to be epistemologically fruitful. The third point, intertwined with and resultant of the first two, emphasizes that inquiring into the right occasioning factor was a dialectical endeavour whereby anticipating potential objections—and responding to real-time objections—might ground the choice of one occasioning factor over another, suggest an alternative one, or assist in selecting one from among several competing claims. Finally, the fourth point is simply that such dialectical inquiries were dynamic, in the sense that one’s conclusion could be revised in view of counterarguments introduced during the dialogical exchange. In what follows, we will elaborate on these points.

Istiqrāʾ and Abduction

both are ampliative, meaning that the conclusion goes beyond what is (logically) contained in the premises (which is why they are non-necessary inferences), but in abduction there is an implicit or explicit appeal to explanatory considerations, whereas in induction there is not; in induction, there is only an appeal to observed frequencies or statistics.

Thus, inductive inferences “may be characterized as those inferences that are based purely on statistical data, such as observed frequencies of occurrences of a particular feature in a given population” (Douven 2021).

• Most 11^th-12^th c. CE Muslim thinkers wrote primarily in Arabic;

• al-Ghazālī was an 11^th-12^th c. CE Muslim thinker;

• Therefore, al-Ghazālī wrote primarily in Arabic.

Importantly, this does not show why most 11^th-12^th c. CE Muslim thinkers wrote primarily in Arabic. The frequent occurrence of Muslim thinkers who wrote primarily in Arabic does not prove, but—to borrow the words of Hallaq (1990, p. 3-5)—it lends corroborative support to the first premise. And, in so doing, it suggests that the best way to distinguish between induction and abduction is precisely as Douven has it in the citation above: “in abduction there is an implicit or explicit appeal to explanatory considerations, whereas in induction there is not; in induction, there is only an appeal to observed frequencies or statistics.”

46It is clear that in juristic contexts carrying out a survey in order to establish a universal law by means of istiqrāʾ requires examining sources and cases which are, in principle, quite different, but nevertheless suitable to be subsumed under the same law. In consequence, the explanatory hypothesis for the examined cases’ causation will appear to shape the generalization underlying this kind of istiqrāʾ. Furthermore, when we shift from legal to natural causation, this explanatory element in istiqrāʾ will be presupposed, but, in our view, not sufficiently stressed.

47A common example of juristic istiqrāʾ, as discussed by Hallaq (1990, p. 6-7, citing al-Ghazālī among others), is the case of the non-obligatory payer known as watr. The argument might be formulated as follows:

• No obligatory prayers examined in our survey are allowed to be performed while on a journey;

• Watr is allowed to be performed while on a journey;

• Therefore, watr—despite not being mentioned in the survey—is recommended, not obligatory.

As for istiqrāʾ, it is an expression which denotes the examination (taṣaffuḥ) of particular things [cases] so that we might judge by their ruling for something which includes those particulars.

It is like our saying, with regard to watr: “It is not obligatory, because it is performed on a journey, and the obligatory [prayer] is not performed on a journey.” Then it is said: “Why did you say that the obligatory [prayer] is not performed on a journey?” So, it is said: “We know it by way of istiqrāʾ, since we saw that the make-up performance [of a prayer] (qaḍāʾ), and the on-time performance (adāʾ), and the subject of a vow (mandhūr), and the rest of the types of obligatory [prayers] are not performed on a journey; thus, we said: All obligatory [prayers] are not performed on a journey.”

The manner of indication of this is not completed except by the first arrangement [i.e., the first figure], in that we will say: “Every obligatory [prayer] is either an on-time performance, a make-up performance, or a vow; and every make-up performance, on-time performance, and vow is not performed on a journey; therefore, Every obligatory [prayer] is not performed on a journey.”

But this is something uncertain (mukhīl), which is right for propositions of preponderant belief (ẓanniyyāt) but not propositions of certainty (qaṭʿiyyāt). The shortcoming is under his saying “either an on-time performance,” for his ruling that “every on-time performance is not performed on a journey” will be denied by the opponent—since watr, according to him, is an obligatory performance, and is performed on a journey.

Of the on-time performances, the opponent will concede only the five prayers, and this is a sixth prayer, according to him, so he will say: “Did you examine (hal istaqrayta) the ruling of watr in your examination? And how did you find it?” Then if you say: “I found that it is not performed on a journey,” then the opponent will not concede. And if you did not examine it, then nothing but some of the on-time performances are evident to you, so the second premise departs from being general and becomes specific. But that does not conclude, because we have explained that the second premise in the first arrangement has to be general.

Due to this, the one who says: “The Maker of the world is a body” has erred (ghalaṭa), because he says: “Every agent is a body; and the Maker of the world is an agent; therefore, He is a body.” Then it is said: “Why did you say ‘Every agent is a body’?” So he will say: “Because I examined the agents among tailors, masons, shoemakers, cuppers, blacksmiths, and others, and I found them all to be bodies.” And it will be said: “And did you examine the Maker of the world, or not? For if you did not examine Him, then you have examined some but not all, and found some of the agents to be bodies, so the second premise has become specific and does not conclude. But if you examined the Creator, then how did you find Him? For if you say: ‘I found Him to be a body,’ then this is the very point of contention—so how have you inserted it into the premise?”

By this it is affirmed that if the istiqrāʾ is complete, then it reverts to the first arrangement [i.e., the first figure], and is right for propositions of certainty (qaṭʿiyyāt). But if it is not complete, then it is right for nothing but juristic propositions (fiqhiyyāt), because whenever the greater part is found to be in a particular manner, it becomes an overwhelmingly preponderant belief (ghalaba ʿalā al-ẓann) that the others are likewise.

49With istiqrāʾ providing the grounds for concomitance, it would seem that the justification of a causal link between a claimed madār’s existence and its corresponding dāʾir’s existence is corroborated by affirming the universal law suggested by the dawarān. However, this corroboration is (almost?) never complete, which strongly suggests that istiqrāʾ can only ever be a part of any larger endeavour of causal justification, which, as we have argued above, may involve the whole gamut of dialectical means for determining the ʿilla.

Arguments, Rather than Inferences, to the Best Explanation

Given evidence E and candidate explanations H₁,…, H_n of E, if H_i explains E better than any of the other hypotheses, infer that H_i is closer to the truth than any of the other hypotheses.

As Douven next points out, this formulation “clearly… requires an account of closeness to the truth, but many such accounts are on offer today.”

51As for the context of Islamic legal theory and dialectic, however, we might observe the following:

• • 43 See, e.g., the tarjīḥāt, the method of istiḥsān, and other modes mentioned above in the introductio (...)

  Degrees of “closeness to the truth” were achieved by confronting competing solutions with series of criteria for preponderance and preference, which were established by jurists in consideration of manifold sources.43 (Note that in consequence, if we wish to apply similar preponderance and preference criteria to natural causation claims, then a suitable extension of the legal criteria must be developed.)

• • 44 See, e.g., the qiyās-relevant mode of istidlāl bi-l-taqsīm, and the dialectical objection of muʿāra (...)

  Pools of competing solutions were often generated by defining legitimate dialectical objections and responses which could govern rebuttal and further dialectical interaction until some agreement was settled. (Note that competing solutions might propose the same conclusion while also providing better explanations for a ruling’s establishment—i.e., a more plausible occasioning factor44—or they might indicate that not only does the claimed property not occasion the claimed ruling, but it occasions another ruling wholly incompatible to that which is claimed.)

These two observations suggest that, in the context of Islamic legal reasoning, we should rather speak of arguments (rather than inferences) to the best explanation, thus evoking a dynamic picture allowing the display of whole structures of sub-arguments, sensitive to changes, at the play-level.

52Perhaps the simplest articulation is to say that the larger project of causal justification in Islamic legal theory and dialectic has a double nature; namely:

1. As a set of justification procedures; and

2. As the result of such procedures.

In the first sense it is an act and in the second an object. As an act, it is a set of dialectical processes involving epistemological, hermeneutical, logical, and legal inquiries which can prompt sub-arguments based on objections and responses, counter-objections and counter-responses, until the matter is settled. And when that matter is settled, an object is produced: the justification, as a piece of evidence.

54From a dialectical stance, on the other hand, it is the process that settles the justification of each of the premises placed under scrutiny. It is not the justified implication or inference that is defeasible or non-monotonic, but, rather, it is the justification process that is the target of possible revision, endorsement, or re-assessment. And this process has its own dynamic, regulated by a dialectical framework within which—at least in the legal realm—preference and preponderance criteria guide us in drawing the best choice so far as knowledge of the contended subject goes; and, importantly, these criteria are also the subject of constant scrutiny. Be that as it may, and leaving contemporary, non-monotonic logicians to deal with their own worries and means, the double nature of causal justification in Islamic legal theory, so we suggest, offers a novel, patently dialectical approach to what should be called argument (rather than inference) to the best explanation.

Conclusion and Work Ahead

1. a gamut of procedures for justifying the choice of one argument over another; and

2. a transference procedure of parallel reasoning that allows the extension of a ruling’s scope of application to new cases not yet integrated into the body of the Law.

The second feature of this novel form of abduction, i.e., the transference procedure, includes an important difference with respect to contemporary approaches to abduction; namely, the fact that the ʿilla and ḥukm of the root-case (the explanandum) might be refined or reassessed (e.g., by istidlāl bi-l-taqsīm, or the objection of farq), or even supplanted in favour of an exception (by istiḥsān). The dialectical form of abduction at work in juristic qiyās is more general than contemporary approaches precisely because the explanandum might remain an object of further scrutiny, in relation to a continued search for appropriate explanatory premisses. Moreover, such justification and transference procedures require that the overall operation be made explicit, and that the efficient process (or function) occasioning presences of the ruling from presences of the property (or properties) be identified.

56Returning now to dawarān, we may therefore say that it prescribes exactly that one examines if the claimed presences of the madār are concomitant with the ruling whose ʿilla is the object of inquiry. Indeed, if we recall our previous analysis, it suggests that a kind of implication, or even some form of bi-implication, is involved:

57But let us shift focus to the challenges one will face if the efficient causal process is left implicit or thought to be encoded in the madār:

X ! (∀x: madār) dāʾir(x) ∧ (∀y: ~madār) ~dāʾir(y),

[NB: the exclamation mark indicates that interlocutor X affirms that (s)he is in possession of some method (based on ṭard and ʿaks) for justifying the conjunction—but without making this method explicit.]

Note this does not allow us to question the exact process claimed by X to justify the assertion. Moreover, if we take seriously the fact that mere concomitance is not sufficient, we will require additional dialectical procedures to corroborate the causation claim—and, also, to give it preponderance over alternative arguments—and this will require that the deployed justification procedure be made explicit.

58But instead, let us go a step further towards a simplified concept of dawarān, and present it as the conjunction of two implications with implicit domain, as in standard, first-order logic:

X ! ∀x madār(x)⊃ dāʾir(x) ∧ ∀y: ~madār(x) ⊃ ~dāʾir(y)

Note that if the madār is multiple—that is, there is a compound of concomitant properties such that some (e.g., the smell or colour of wine) are inseparable from those which are efficient—then the above analysis, which leaves the existence of tokens of the claimed madār tacit, is quite unhelpful.

59What we need for the analysis of such cases is not only a means by which to record the concomitance of a property’s existence with its claimed legal effect, but also a means by which to distinguish the causal contribution of each of the components—particularly when these components constitute a mereological whole:

60Alternatively, if we express this as an implication, even one with a tacit efficient cause, an analysis aiming to identify the relevant compound will indicate, in our example, that the interdiction for consumption has, as scope, all instances of intoxicating drinks:

• 49 See Shahid Rahman & Mohammad Saleh Zarepour (2021), “On Descriptional Propositions in Ibn Sīnā”, p. (...)

X ! {∀z: (x: Intoxicating-Drink | Smell(x))} Forbidden(fst(z))49

1. compound antecedents, as with the multiple / compound madārs mentioned above;

2. integrating the dynamic of the dialectical account of causation—including a theory of argument to the best explanation; and

3. avoiding the reduction of causal to purely analytical assertions.

64According to our view, this grasping of the universal is prompted, in the legal realm, by the various abductive methods found within the dialectical framework of qiyās, and especially its larger, critical project of causal justification. The key point here is that the legal framework provides a theory allowing one to identify the best causation argument from among competing arguments. Occasionally this will even yield certain (qaṭʿī) knowledge (for example, when the cause is directly identified in revelation), but mostly it will not. However, even in the latter case, the diverse dialectical methods will yield a “preponderant belief” (ẓannī) level of knowledge which can be taken as most likely vis-à-vis the available justification and testing methods.

65However, if such a methodology is to be transferred from the legal to the natural realm, then we will need, among other things, some efficient instruments for choosing the most likely explanation. At this point in our research, we cannot yet say how this could be (or was?) achieved, though the pathway likely involves appeals to certain metaphysical, as well as epistemological, principles. This suggests that quite a large amount of work lies ahead; but we are eager to further explore the avenues which Islamic legal theoretical and dialectical traditions have opened with regard to causal justification, and welcome further comment and collaboration.

Appendix: al-Ghazālī on the inutility of concomitance as a test for causation

66Sources:

67[ب] = [B] : al-Ghazālī, Al-Mustaṣfā, Būlāq ed. 2.307-9

68[ح] = [Ḥ] : al-Ghazālī, Al-Mustaṣfā, Ḥāfiẓ ed., 3.635-9