Avicenna's Notion of Fiṭrīyāt: A Comment on Dimitri Gutas' Interpretation

I. Introduction

In an illuminating article, Dimitri Gutas has tried to show that Avicenna's theory of knowledge should be understood within a full-blown empiricist framework very similar to that of John Locke.¹ Gutas' argument is based on an analysis of Avicennian 'principles of syllogism'² (mabādī al-qiyās). The principles of syllogism are those judgments and propositions that form the irreducible and axiomatic foundations of syllogisms and definitions.³ Avicenna categorizes these principles based on how we accept and acknowledge the truth (taṣdīq) of them. This categorization appears, with some slight modifications, in many places in Avicenna's oeuvre, for example in the Kitāb al-Burhān of al-Šifā',⁴ and the logic parts of al-Nağāt⁵ and al-Išārāt wa-l-tanbīhāt.⁶ According to al-Nağāt, the principles of syllogism are divided into sixteen types based on the cognitive mechanisms through which we grasp them.⁷

Gutas' argument for his main claim has two steps: (a) he considers these different types of principles, one by one, quoting the most important texts in which Avicenna discusses each of these principles; and (b) he shows, based on his analysis of the selected quotations, that the acknowledgment of the truth of these principles is finally grounded on what we grasp from our sensory experiences (mušāhadāt). More precisely, Gutas wants to show, based on the textual evidence, that we cannot acknowledge the truth of the principles of syllogism merely by the operation of our theoretical intellect and without appealing to what we obtain non–a priori, in the Kantian sense.⁸ The conclusion is that, according to Avicenna, all sorts of human propositional knowledge have an empiricist bedrock.

No one can deny that Gutas' article, like all his other works, includes many valuable and insightful ideas. Despite this fact, I am not completely convinced by his analysis about at least some of the aforementioned principles. Specifically, what he says about the nature of the data with builtin syllogisms (qaḍāyā qiyāsātuhā ma'ahā or muqaddamāt fiṭrīyāt al-qiyās—henceforth fiṭrīyāt) seems implausible to me. Gutas believes that Avicennian fiṭrīyāt are non–a priori and analytic propositions. However, in the present article I argue that a deeper look at Avicenna's writings on fiṭrīyāt reveals a plausible view that is very different from Gutas' position. I try to show that, contrary to what he proposes, Avicennian fiṭrīyāt are synthetic a priori.

My article is organized as follows. The next section below presents the details of Gutas' views about fiṭrīyāt. I show that there is an equivocation in the term 'fiṭrīyāt' and that it can be understood in at least three different ways (when it is employed as a property for judgments). In section III, I show that at least some fiṭrīyāt are a priori judgments. In section IV, I show, based on Kant's principal criterion for analyticity, that fiṭrīyāt are, contrary to what Gutas explicitly claims, synthetic. Section V concludes.

II. Gutas' Interpretation of Avicennian Fiṭrīyāt

Gutas argues that primary data and fiṭrīyāt are two groups of principles of syllogism "whose necessity is internal to the soul and is imposed by the intellect."⁹ The difference between them is that the truth of the former group is directly imposed by the intellect, while the truth of the latter group is only indirectly imposed by the intellect, that is, "through an operation natural" to it.¹⁰ This natural operation is the fiṭra of the intellect. In fact, each of our cognitive faculties, Gutas argues, has a natural mode of operation that is its fiṭra. Criticizing the common understanding of fiṭra according to which fiṭra is the main source of innate knowledge, Gutas tries to show that Avicennian fiṭra is not an independent cognitive faculty;¹¹ it is merely the natural mode of the operation of our cognitive faculties.¹²

Moreover, Gutas emphasizes that Avicenna "denies innate ideas. His theory of the rational soul, well studied by now, is unambiguous. Upon birth, the newly created intellect that is associated with the body is absolutely potential, a tabula rasa."¹³ In Avicennian epistemology, Gutas believes, "[t]he human rational soul, upon its first creation and association with the body of the newly born infant is absolutely potential, a tabula rasa. … As the child grows up and until it reaches maturity…, Experience (mušāhada) provides him with information about the world and himself."¹⁴ He finally concludes that "[t]he function of the intellect is procedural; in itself it has no innate or a priori contents."¹⁵ Gutas argues that in Avicenna's theory of knowledge there is no room for either innate propositional knowledge (al-ilm altaṣ dīqī) or innate concepts (al-ilm al-taṣawwurī). It is worth mentioning that some pieces of our propositional knowledge, for example self-awareness, and some of our concepts, for example the concept of existence, have been argued by some scholars of Avicenna's philosophy to be examples of innate knowledge.¹⁶ Yet Gutas believes that they are not really pre-given, inborn, or innate: all of them are grounded finally on what we obtain from our experiences of the external world (by sense perception) and/or our own internal world (by reflection).

Gutas quotes the following passage from al-Nağāt as evidence in support of the aforementioned views about fiṭrīyāt:

[The floating man can submit something to his mind and raise] a doubt about it. If he is able to doubt it, then his fiṭra does not attest to it; but if he is not able to doubt it, then it is something which his fiṭra imposes. But not everything which the human fiṭra imposes is true, but many of them are false. True is only the fiṭra of the faculty called intellect. … [Sometimes the fiṭra of estimation makes wrong judgments].¹⁷

Only judgments and propositions (not concepts) can be considered as fiṭrī in the sense expressed in this passage. This is because concepts have no truthvalue and cannot be true or false. The second point that can be understood from this passage is that all fiṭrī judgments have some kind of psychological (or phenomenal) necessity; that is, they cannot be denied by anybody who considers them. If we accept that logical necessities are the only propositions that can neither be false nor be conceived to be false, then it follows from the passage above that the judgments imposed by the fiṭra of the intellect are logically necessary.¹⁸ By contrast, the judgments imposed by the fiṭra of the other cognitive faculties (e.g., judgments imposed by the fiṭra of the estimative faculty), though seemingly true, may be false.

This shows that there is an equivocation in the term 'fiṭrīyāt'. It seems that it can be understood in at least three different ways (when it is employed to describe a property of some judgments). According to its most general meaning, every judgment imposed by the fiṭra of one of our cognitive faculties is one of fiṭrīyāt. According to the second meaning, which is less general, only those judgments imposed by the fiṭra of the intellect can be considered as fiṭrīyāt. In this case, not only data with built-in syllogisms but also primary data (see endnote 7) should be considered as fiṭrīyāt. This is because both of them are imposed by the fiṭra of the intellect.¹⁹ But what I refer to by the term 'fiṭrīyāt' in this article is restricted only to the data with built-in syllogisms. This is the third and the narrowest meaning of the term 'fiṭrīyāt' that we may have in interpreting Avicenna. As we will see, confusing these different understandings of fiṭrīyāt may lead us to an inaccurate analysis of the epistemic and semantic status of fiṭrīyāt.

What, then, is the difference between primary data and fiṭrīyāt (in the sense that we use this term) for Avicenna? Gutas argues that the difference is only that every primary datum can be verified just by understanding its terms (i.e., its subject/predicate or minor/major terms) without any need to construct a syllogism. But each fiṭrīyāt can be verified only after constructing a syllogism whose middle term will automatically appear in the mind, immediately after understanding the terms of that proposition.²⁰ Gutas does not believe that this difference plays any important epistemic or semantic role. He thinks that primary data and fiṭrīyāt have a completely similar epistemic and semantic status: both of them are non–a priori and "analytic, in Kantian terms."²¹ In the next two sections I will argue that this understanding of Avicenna is not tenable. More explicitly, I think both of these groups of principles of syllogism are a priori. However, unlike primary data (which are analytic), fiṭrīyāt are synthetic.²²

III. Avicennian Fiṭrīyāt Are A Priori

I think that Gutas' understanding of the notion of a priority does not correspond to the standard interpretation of this notion (at least in its Kantian sense). More precisely, it seems to me that two things are problematic with his understanding of this notion. Let me explain.

First, he thinks that the notion of a priority is identical to the notions of innateness and pre-givenness. Therefore, he thinks that the sufficient condition for the non–a priority of a concept or a judgment is for it to be not innate or given inborn. 'A priori' for Gutas means "what we have in our soul prior to our birth." For example, he says: "In all [of its] operations the function of the intellect is procedural; in itself it has no innate or a priori contents."²³ Moreover, in order to argue for the non–a priority of principles of syllogism, Gutas refers (as his evidence) to the texts in which Avicenna says that these principles "come about" in the human intellect as a man grows up.²⁴ But as we will see, this picture of the connection between these notions (i.e., innateness and a priority) is oversimplified.

Second, Gutas reduces the a priority of a judgment or proposition to the a priority of the concepts from which that proposition is constituted. In other words, he considers every proposition constituted from non–a priori concepts to be a non–a priori proposition. For example, in order to show that the proposition 'the whole is greater than the part' (as a primary datum) is non–a priori, he argues just that the concepts 'the whole', 'greater', and 'part' come, through abstraction, from sensation.²⁵ Again, the relation between the a priority of a proposition and the a priority of its conceptual components seems to be more complicated than what we see in Gutas' picture. In the remaining of this section, I will clarify Kant's view on the notion of a priority. Based on that, I will show that both primary data and fiṭrīyāt can be considered as Kantian a priori judgments.

Kant uses the notion of a priority as a property for at least three different things: concepts, judgments, and justifications/warrants for judgments.²⁶ A priority is the property of being independent of all experiences and even of all impressions of the senses. An a priori concept is formed independently of all our experiences and impressions from the external world. A concept is a priori if it is in principle possible to have and entertain this concept without having any experience of the external world. According to this understanding of the a priority of concepts, a concept may be a priori but not innate. However, all innate concepts are a priori. Kant without reservation believes in innate concepts—for example time, space, and the categories.²⁷ But he also believes in some other a priori concepts that are not innate. For example, the concepts of numbers are formed "through successive addition of units in time."²⁸ The concept of 'two' therefore arises from the a priori intuition of two successive units in time. It is not pre-given and inborn; it comes about in our intellect as we grow up. Nonetheless, Kant considers this concept to be a priori, since what we grasp from our experiences of the external world has no significant role in its formation. This analysis clearly shows that in order to prove the non–a priority of a concept it is not sufficient to argue that it is not innate and that it comes about in our mind as we grow up. A concept is non–a priori only if it is formed by what we obtain from our experiences of the external world. Consequently, even if, according to Avicenna, we accept that there is no innate concept, this does not necessarily mean that all concepts are non–a priori for him.²⁹ Now, does Avicenna believe that all concepts are non–a priori in the aforementioned Kantian sense? And even if he does, does Gutas provide sufficient reasons to show this?

Gutas does not provide any general argument to show that all concepts are a priori according to Avicenna. As I mentioned above, Gutas has argued that concepts such as 'the whole', 'greater', and 'part' come, through abstraction, from our experiences of the external world. So if his argument is sound, it entails that these concepts are non–a priori. But he has not shown that concepts like 'four' and 'even' are formed through the mediation of our sense-perceptual experiences; he has only said that they 'come about' in our mind as we grow up. This is definitely insufficient to show that these concepts are non–a priori in its Kantian sense.³⁰ Coming-about-in-mind is not the same notion as non–a priority. Moreover, I think that if we accept the condition of being independent of sense-perceptual experience as the criterion of a priority (instead of the innateness or not-coming-about criterion), then we should conclude that Marmura is right about the a priority of some primary universal concepts like 'necessity', 'thingness', and 'existence' in the framework of Avicennian philosophy.³¹

The problem goes deeper. Even if we accept that all concepts, including all conceptual components of primary data and fiṭrīyāt, are non–a priori, we cannot still conclude that these principles are non–a priori judgments. To clarify Kant's position on the a priority of judgments, we should first discuss the a priority of justifications. A justification for the truth of a judgment is a priori only if it is not based on empirical evidence that cannot be grasped without having some experience of the external world. Such a justification cannot be defeated by empirical information. Consider that a person S has a justification J for the truth of a judgment P. J is a priori only if its validity and plausibility are independent of the condition of the possible worlds that S inhabits.³² The a priority of J is completely independent of the a priority of the conceptual components of P. It is possible for P to be justified by an a priori justification even though it contains some non–a priori concepts. Consider the proposition 'bachelors are unmarried'. This is constituted from the concepts 'bachelor' and 'unmarried', which are non–a priori concepts since both are formed based on our experiences. But its justification is a priori, since after grasping and understanding its conceptual components, one would not need empirical information to acknowledge the truth of this proposition. That the typical example of propositions with a priori judgments—that is, 'bachelors are unmarried'—is constituted from non–a priori concepts shows how crucial it is not to confuse the a priority of concepts with that of justifications.

Here, again, the innateness of a judgment cannot be concluded from the a priority of its justification. Kant believes that all plausible justifications of all mathematical judgments are (and must be) a priori; but he surely does not believe that mathematical judgments are innate. As a result, that a judgment is not innate does not entail that it has a non–a priori justification. It indicates that the a priority of the justifications of primary data and fiṭrīyāt cannot be rejected by the mere fact that these propositions are not innate and pre-given. Quite the contrary, it seems that if we endorse the aforementioned criterion for the a priority of justifications, then we should say that both Avicennian primary data and fiṭrīyāt have a priori justifications. In order to acknowledge the truth of these principles no empirical information is required. After grasping the conceptual components of these judgments, we can acknowledge their truth without the aid of senseperceptual experiences. In his discussion of the fiṭrī proposition 'four is even' Avicenna argues that after grasping the concepts 'four' and 'even' we do not need any further thing (which a fortiori includes any extra empirical information) to acknowledge the truth of the proposition that four is even: "It will be represented to everyone who understands [the concepts] 'four' and 'even' that four is even."³³ Assenting to the truth of fiṭrīyāt is internal to the intellect and does not depend on the empirical information we grasp through our senses. So it seems that, contrary to what Gutas suggests, according to Avicenna fiṭrīyāt are justified in an a priori way.

Now, one might object that the a priority of justifications and the a priority of judgments are two different things, and that Gutas has discussed the latter but not the former. So we should clarify Kant's position on the a priority of judgments. According to Kant, we have three different kinds of a priori judgments. All of them have a priori justifications. Therefore, it is not possible for a judgment to be a priori without having an a priori justification. The first group of a priori judgments are constituted from innate concepts (e.g., time, space, the categories). They are pure a priori judgments that form the categorical bases of our experiences. The second group of a priori judgments are constituted from a priori, but not innate, concepts. Mathematical judgments are the best-known representatives of this group.³⁴ The third group of a priori judgments are constituted from non–a priori concepts (but, as I said, they have a priori justifications). The proposition 'bachelors are unmarried' is an example of such judgments. These are impure a priori judgments.³⁵

It should be noted that Kant scholars insist that he "agrees with Locke that we have no innate knowledge, that is, no knowledge of any particular propositions implanted in us by God or nature prior to the commencement of our individual experience."³⁶ Therefore, the first group of aforementioned a priori judgments are surely not declarative propositions that acknowledge the truth of something. They just form a categorical ground or framework for understanding the world through experience.³⁷ This means that in the discussion of the a priority of declarative judgments—for example in the discussion of the a priority of Avicennian principles of syllogism—we should restrict ourselves to the last two groups of propositions. In this sense, it is very difficult to deny that Avicennian primary data and fiṭrīyāt are a priori. These principles, though possibly constituted from non-innate concepts, are a priori, since they have a priori justification. Now then, we can discuss the analyticity of Avicennian primary data and fiṭrīyāt.³⁸

IV. Avicennian Fiṭrīyāt Are Synthetic

Gutas believes that "both kinds of propositions [i.e., primary data and fiṭrīyāt] would be analytic, in Kantian terms."³⁹ I disagree. There is some evidence that Kant proposed three different definitions for the notion of analyticity.⁴⁰ One of them is based on the notion of containment, the second one on contradiction, and the last one on amplification. According to the first definition, a thought is analytic if and only if its predicate is contained in its subject. According to the second definition, a thought is analytic if and only if its negation is contradictory. Finally, according to the third definition, a thought is analytic if it is an amplification (not only a clarification). Discussing these definitions and clarifying their principal differences is beyond the scope of this article. Here, I rely on the first definition, since it has been argued, convincingly in my opinion, that it is more fundamental for Kant and that the two other definitions finally reduce to this definition.⁴¹ Therefore my criterion for evaluating the analytic status of Avicennian primary data and fiṭrīyāt would be the following:

(D) A subject-predicate judgment is analytic if and only if its predicate is contained in its subject. A subject-predicate judgment is synthetic if and only if it is not analytic.

This definition is based on the notions of 'subject' and 'predicate'. But it can easily be paraphrased in another way, more familiar to Aristotelian-Avicennian logicians:

(D*) A judgment is analytic if and only if its major term is contained in its minor term. A judgment is synthetic if and only if it is not analytic.

Now, based on (D*), we can see that primary data are analytic, and fiṭrīyāt, contrary to Gutas' proposal, are synthetic. Gutas says that "[t]he difference between these propositions with built-in syllogisms and the primary ones mentioned above is that the former, according to Avicenna, can be verified by a syllogism whereas the latter require only understanding the terms of the propositions and their combination."⁴² In other words, a primary judgment has no middle term but a fiṭrī judgment has a middle term. If a person considers a primary judgment, as soon as he understands its term he will find out that the major term is contained in the minor term. Therefore he can immediately acknowledge the truth of this proposition. This means that, based on (D*), primary data should be considered as analytic propositions. On the other hand, understanding the terms of a fiṭrī judgment is not sufficient for acknowledging its truth, as its major term is not contained in its minor term: it has a middle term. Of course, if one considers a fiṭrī judgment, immediately after understanding the terms of the judgment the middle term will appear in her mind by the fiṭra of her intellect. Then she can immediately acknowledge the truth of that proposition. But the major term of a fiṭrī judgment is not contained in its minor term, and this is sufficient for them to be synthetic, based on (D*).

V. Conclusion

An oversimplified picture of the Kantian notions of a priority and analyticity may take us into adopting an inaccurate view about the status of fiṭrīyāt. Regarding the issue of a priority, I showed the following: (1) A priority and innateness are two distinct notions. Every innate concept or judgment is a priori, but not vice versa. Therefore, the non–a priority of a concept or judgment cannot be concluded from its non-innateness. (2) The a priority of a judgment is entirely independent from the a priority of its conceptual components; neither entails the other. Consequently, even if we accept that fiṭrīyāt are not innate and are constituted from non–a priori concepts, we cannot conclude that they are non–a priori judgments. In contrast to Gutas' view, fiṭrīyāt are a priori, since after grasping their conceptual components we can acknowledge their truth without the need to appeal to our sense perceptions of the external world. In other words, although the conceptual components of fiṭrīyāt may be formed non–a priori, their justification is a priori. On the other hand, fiṭrīyāt are synthetic, since they have a middle term and their major term is not contained in their minor term. Hence, Avicennian fiṭrīyāt are synthetic a priori in the Kantian sense.

The disagreement between empiricism and rationalism is not restricted to the issue of whether or not we have pieces of innate knowledge. The belief that we have innate and pre-given knowledge is sufficient but not necessary for being non-empiricist or rationalist. The real challenge between these two camps lies in the issue of how to justify (or acknowledge the truth of) judgments. Empiricists believe that no non-analytic judgment (even after forming or grasping all of its conceptual components) can be justified a priori. This is what we must show regarding Avicenna's epistemology if we want to tag him as an empiricist.