1. Which Question Does the Flying Man Address?

Adamson and Benevich claim that in Shifāʾ: Nafs I.1, Avicenna is dissatisfied with the Aristotelian definition of the soul as ‘the first perfection of a natural organized body that performs acts of life’ (Avicenna, Shifāʾ: Nafs I.1, 12; cf. Ar. De an. II.1, 412a27–28; unless mentioned otherwise, all translations from the Arabic are mine). According to them, the flying man argument should be understood as an attempt at a more adequate formula, in particular, one that does not include the relation to the body, allegedly an accidental feature, in the definition of soul (Adamson and Benevich Reference Adamson and Benevich2018: 152–53; for a similar argument, see Alpina Reference Alpina2018: 191–202).

It is difficult to reconcile this claim with Avicenna's method of procedure in the chapter as a whole. He begins from the observation of phenomena specific to living things, and concludes that since not all bodies exhibit these phenomena, they cannot be due to corporeality as such. Thus, a further principle is needed, and he says:

Here, at the very beginning of the chapter, Avicenna clearly states that ‘soul’ is a relational term that designates the thing that is a principle of life insofar as it is a principle of life, that is, insofar as it is related to living bodies. Today we would perhaps say that ‘soul’ is a functional term. He explicitly says that ‘soul’ remains neutral about what the thing that is such a principle is in itself, and he adds that this will be dealt with later. Whether this means later on in the same chapter, later on in psychology, or perhaps even later on in the philosophical curriculum, is left unspecified.

When it comes to this relative or functional concept of soul, however, Avicenna happily endorses Aristotle's definition. This is corroborated by his evaluation of the different candidates for the definition of soul. He first rules out quwwa, that is, ‘power’ or ‘faculty’, because this term signifies both active and passive capacities (for instance, motion and perception, respectively). It is therefore ambiguous, and as an ambiguous term, it is unfit for a real definition. On the other hand, neither of the disambiguated meanings alone is capable of encompassing all life activities, which is why quwwa fails to signify ‘soul absolutely’, that is, all kinds of soul (Avicenna, Shifāʾ: Nafs I.1, 7–8). But pace Adamson and Benevich, Avicenna finds no similar qualms with defining soul as a perfection. Indeed, he has earlier stated unequivocally that ‘it is clear from this, then, that when we define the soul as a perfection, this most properly denotes its meaning and likewise includes all the species of the soul in all respects, not excluding the soul that is separate from matter’ (Avicenna, Shifāʾ: Nafs I.1, 7). The definition of soul as perfection does precisely what the definition as quwwa fails to do: it encompasses all kinds of soul (plant, animal, and human) and thus all life activities. What is more, as Avicenna explicitly mentions, this definition is neutral about the ontological status of the kinds of things that can function as souls—in particular, whether they are forms, which are inseparable parts of hylomorphic compounds, or incorporeal substances (see the long argument in Avicenna, Shifāʾ: Nafs I.1, 8–10).

The slightly later passage that Adamson and Benevich introduce as evidence for Avicenna's rejection of ‘perfection’ as definiens of ‘soul’ is readily explained in this light. Let us recall the passage:

This is not a rejection of the Aristotelian definition of soul, but a qualifying remark, according to which the definition only holds for the functional term, remaining neutral about the ontological status and the constitutive features of the things that function as souls. Avicenna hastens to add that this is the concept of soul proper to natural philosophy, of which psychology is a branch, precisely because the concept includes ‘a connection to matter and motion’ (Avicenna, Shifāʾ: Nafs I.1, 11). It is true that Avicenna elsewhere uses the term ‘soul’ to refer to the thing that functions as a soul, seemingly independent of this function of animating the body—think about phrases like ‘separate soul’ (nafs mujarrad) or ‘rational soul’ (nafs nāṭiqa). However, I do not think this is evidence for Avicenna rejecting the Aristotelian definition as improper for the science of psychology, the subject matter of which is constituted by the functions of life ranging from the most basic vegetative operations to the most sophisticated levels of cognition. At best, Avicenna's use of such phrases tells us that ‘soul’ can be said in different senses, and it is perfectly reasonable to assume that Avicenna would have held the looser use to be ultimately grounded in the scientific one, as shorthand for ‘the thing that functions as a soul’.

On the heels of these considerations, I find it unlikely that the flying man argument is concerned with the quiddity of soul, simply because that quiddity has already been acquired. In other words, the argument is not concerned with the constitutive features of being a soul in the absolute sense of the word—we already know what those features are. What, then, is the argument about? Heeding the example of Adamson and Benevich and taking our cue from Avicenna's logic, we find that in Shifāʾ: Burhān I.5 (references to this title are to ʿAffīfī's [Reference ʿAffīfī1956] edition), Avicenna elaborates on the Aristotelian distinction between the different kinds of scientific questions (cf. An. post. II.1) as well as on the related theory concerning their proper order in scientific inquiry. Once we have a nominal definition for a scientific term, the first question to ask is whether (hal) anything instantiating the nominal definition exists in the extramental world. This is the starting point of all science insofar as a positive answer to the question whether establishes that there is something for the science to study in the first place. The question whether is in turn divided into two types, namely, the simple (basīṭ) and the composite (murakkab) whether. The simple one asks whether something denoted by the subject term exists pure and simple, whereas the composite one asks whether something exists as something or under a certain qualification. Finally, once we have located grounds for a positive answer to a whether question of either kind, we can proceed to ask anew what (mā) that existing thing is. This second what is not a mere repetition of the initial question, for this time we are not inquiring about the meaning of the subject term but about the real definition of the thing that is called by the term and that has now been shown to exist. A successful answer to the second question what (the real what as opposed to the first, nominal what) is a detailed account of the constitutive features of the thing's quiddity (māhīya) (Avicenna, Shifāʾ: Burhān, I.5, 68–69; a third kind of question is the why [limā], which can only be dealt with once the first two kinds of questions have been answered, but this question is not relevant to our present concern).

Now, Shifāʾ: Nafs I.1 begins with a simple whether question, namely, whether the subject of psychology, or ‘the thing we call soul’, exists.Footnote ¹ This question is answered affirmatively by recourse to the empirical observation of life processes, which although they take place in bodies, are inexplicable by means of the essence of corporeality alone: if life were grounded in corporeality pure and simple, then all bodies would be alive by definition. Hence, an explanatory factor must be added to mere corporeality, and this factor is what we are accustomed to call soul. The chapter then proceeds to ask the real what question about soul and answers it by means of the Aristotelian formula that defines the soul as the first perfection of a natural organized body that is potentially alive. It is at this point, after dealing with all three steps, that we encounter the flying man argument. In light of the foregoing, it is clear that we need not deal with the real what question anymore, let alone ask whether there are souls pure and simple. However, the investigation remains concerning the quiddity of the thing that functions as soul, only now considered in itself and not insofar as it is a soul. In order to get this question off the ground, the first question to ask is whether there is anything to be investigated in the first place, that is, whether the thing that functions as a soul has an existence apart from its being a soul. Thus, it is time for a composite question of whether, namely, the question of whether there are things that function as souls, but exist independently of the bodies they animate.

Attending to Avicenna's introduction of the thought experiment, we find that he is explicit about these methodological points. Let us retranslate the relevant sentences:

There are two things to note about this introduction of the flying man. First, Avicenna explicitly says that we already have the definition of ‘soul’ as a relational, or functional, term. He also tells us that the relational term does not inform us about the quiddity of the thing that functions as a soul, considered in itself and apart from its being a soul. What he does not say, pace Adamson and Benevich, is that we lack a satisfying definition of soul. On the contrary, the investigation that still lies ahead concerns the quiddity of the thing that we already know is a soul—the question of what that thing is in itself, regardless of its function as a soul.

On a related note, Avicenna characterizes the Aristotelian definition here as giving us ‘the meaning of the name’ ‘soul’. This might suggest that Adamson and Benevich are right: so far, we have only acquired a nominal definition and must still nail down the real definition of ‘soul’. Even if this were the case, the point I am making remains: the proper question to ask after a nominal definition is whether anything matching the definition exists, and the flying man that follows is intended to prove the existence of the thing that functions as a soul, not its quiddity.

Second, Avicenna does not say that we now turn to address this what question—all he says is that we have not dealt with this question yet. Moreover, if we look at the very last sentence, he does not say that the flying man is designed to answer the what question either. Instead, the flying man is an argument for the existence of the kind of soul we human beings have.Footnote ³ This is another way of saying that it is an attempt at answering a composite whether question concerning the human soul: does it exist independently of the body?

This interpretation is further corroborated by Avicenna's descriptions of what it is that the flying man would be aware of and that we must therefore assert even in the absence of any knowledge of the body. Avicenna characterizes this variously as ‘the existence of his dhāt [that is, essence or self]’ (Avicenna, Shifāʾ: Nafs I.1, 15 bis), ‘his dhāt as something that exists’ (Avicenna, Shifāʾ: Nafs I.1, 15), or in a later version of the argument, ‘the existence of his annīya’ (Avicenna, Shifāʾ: Nafs V.7, 255; strictly speaking, the technical term annīya means ‘thatness’, and it commonly signifies the fact of existing, but since this would be redundant in the present context, I believe that here we should understand it as denoting the individual instantiation of an essence that one is and hence as synonymous with dhāt). Thus, the single consistent feature here is the emphasis on existence, which is precisely what one would expect of an answer to a whether question. Thus, it is not knowledge of the definition of soul or even of some of the constitutive features of our essence that we acquire but knowledge of the fact that the essence is there, independently of the body. In other words, the flying man provides the basis for studying what the human essence is in itself, apart from its function as a soul. This knowledge can be pursued after the flying man (as Avicenna does in Shifāʾ: Nafs I.3 and V.2), but it is not gained through it.

2. Can the Flying Man Yield Knowledge of Quiddity?

There is another, independent reason for why I find it unlikely that Avicenna designed the flying man to yield knowledge of the quiddity of the thing that functions as a soul in the human body. This is due to Avicenna's general theory of what is involved in a true conception of a quiddity.

As we have already seen, Avicenna explicitly asserts that the flying man is aware of his dhāt. Now, Adamson and Benevich claim that throughout Shifāʾ: Nafs I.1, including in the flying man, the central term dhāt, or ‘essence’, is used interchangeably with māhīya, or ‘quiddity’ (Adamson and Benevich Reference Adamson and Benevich2018: 153). It is true that the two terms are intimately connected, and we can probably find passages from Avicenna where they are used as synonyms. On the other hand, there are equally clear cases where replacing dhāt with māhīya is controversial, such as the sections on self-awareness (shuʿūr bi'l-dhāt) in the Taʿlīqāt (for the relevant texts, see Kaukua Reference Kaukua2015, 55–61). When it comes to our chapter, reading the two terms as simply equivalent seems problematic in the light of passages where Avicenna explicitly distinguishes between them. For instance, very early on in our chapter, he characterizes the order of investigating the quiddity of the thing that functions as a soul in following terms:

Replacing ‘essence’ with ‘quiddity’ would make Avicenna's description of scientific inquiry strangely redundant: once we have verified a quiddity, what could possibly remain that we are not yet acquainted with? But if we distinguish between essence and quiddity, we can understand the verification of essence along the above-described lines as an investigation into whether the essence we are concerned with exists, which then allows us to inquire into the question of what it is, that is, into the constituents and the correct definition of its quiddity. The two terms may refer to the same extramental thing, but they do so in different senses: knowledge of a dhāt is knowledge that the dhāt exists, and this precedes knowledge of what it is, of its quiddity.

This is important because it is the identification of essence with quiddity that allows Adamson and Benevich to claim that the flying man argument is designed to yield knowledge of the quiddity and the constitutive features of soul. Indeed, they claim that ‘[the flying man] is intended to give us a very particular insight about the essence of soul, namely, that this essence requires no connection or relation to body’, or in more positive terms, it ‘helps us to see that it is essential to soul that it be ontologically independent of body’ (Adamson and Benevich Reference Adamson and Benevich2018: 154). In other words, the flying man is designed ‘to ‘remind’ us that we already have a conceptualization of our own souls, which is enough to give us access to the existence conditions of our own souls—in this case, that a connection to body is not an existence condition’ (Adamson and Benevich Reference Adamson and Benevich2018: 162). A negative feature, however, cannot be constitutive of any real entity, and if that is all the flying man argument has taught us, it has not provided us with an insight into the constitutive features of the quiddity of the thing that functions as the human soul. It has only shown us that there are things that function as souls and also exist independent of their bodies, which is an answer to a whether question. From the point of view of the quiddity of such things, their not being connected to bodies is an extrinsic necessary property, or a property concomitant (lāzim) with their quiddities. To put this another way, we can ground the human soul's independence (or more properly, the independence of the thing that functions as a human soul) from its body in the soul's quiddity, but since this is a mere concomitant of that quiddity, it cannot be the basis of our knowledge of the quiddity. In order to get off the ground, the flying man argument of Adamson and Benevich needs to have knowledge of some positive quiddity.

3. The Masked Man and the Principle of Charity

Adamson and Benevich claim that their interpretation has the advantage of being more charitable to Avicenna than existing alternatives (Marmura Reference Marmura1986; Sebti Reference Sebti2000; Black Reference Black, Rahman, Street and Tahiri2008; Kaukua Reference Kaukua2015), with the sole exception of Dag Hasse's similar reconstruction (Hasse Reference Hasse2000: 80–92). This is because their robust account of the content of the flying man's knowledge supposedly renders the argument innocent of the so-called masked man fallacy. Earlier interpreters have attributed to Avicenna a suspicious move from a merely epistemic distinction to a metaphysical one and have thus suggested that he may be guilty of the fallacy. Even if it were the case that I can be aware of myself while unaware of my body, it does not follow that the self is really distinct from the body, just as from my knowing Darth Vader but not knowing Anakin Skywalker, it does not follow that Darth Vader is not Anakin Skywalker. Now, if the flying man does not merely direct our attention to the fact of our self-awareness under a very specific set of conditions, but allows us to conceive of our quiddity veridically, then the inference from the evidence of self-awareness to the incorporeality of our essence is indeed warranted. The argument is no longer a tacit shift from an epistemic distinction to a metaphysical one, but firmly anchored in a metaphysical discussion about the constitutive features of the human essence.

Charitability is a valid virtue in reconstructing the thought of historical authors. But before giving a verdict on the degrees of charity respective to each interpretation, let us consider the cost of the alleged immunity to the masked man. From this point of view, it is important to note that the interpretation of Adamson and Benevich requires the flying man to have knowledge of at least some of the constitutive features of the human essence, despite all the trouble Avicenna takes to convince us that he can perceive, imagine, or think about nothing. If such knowledge is granted, then it is true that the masked man fallacy will not apply, precisely because for Avicenna, ‘human existing in the mind must fully correspond to the extramental essence of human’ (Adamson and Benevich Reference Adamson and Benevich2018: 160). If one can validly think of humanity without thinking of corporeality, then corporeality is not constitutive of humanity.

But why would a reluctant interlocutor of Avicenna—and such there were, for the flying man argument was subjected to explicit criticism by his contemporaries (see Michot Reference Michot1997: 169–74, and Kaukua Reference Kaukua2015: 80–85)—accept that the flying man entertains anything like a mental instantiation of the form of humanity? It is important to notice here that my conceiving the real definition of humanity, or even my conceiving of humanity in some less rigorous sense, is considerably richer in content than, and definitely not the same as, my being aware of myself. I am then thinking of a concept that is an object of a specific kind, albeit one that I can apply to myself, among other things. Hence, in the more robust reading initially put forth by Hasse and now refined by Adamson and Benevich, the argument can no longer hinge on self-awareness alone. But is the flying man's knowledge of himself then anything more than a postulate? If it is not, the flying man is hardly a plausible argument—indeed, it may be debated whether it is an argument in the first place.

Adamson and Benevich seem to recognize this when they write: ‘One may worry here that there is a slip from saying that the flying man is aware of his own ‘essence’ . . . to claiming that the flying man is aware of, or conceptualizes, the essence of his soul. To which we would reply that this is a distinction without a difference: the flying man just is his soul’ (Adamson and Benevich Reference Adamson and Benevich2018: 162). If this means that awareness of one's own dhāt, in the sense of self-awareness, amounts to knowing the essence of soul, in the sense of a conception of its constitutive features, the claim hinges on the ambiguity of the Arabic dhāt, which can mean either ‘self’ or ‘essence’. From a third-person point of view, the flying man's essence (dhāt) is a human essence and will therefore be accurately captured by the concept of humanity once he has acquired it. But why should we believe that his first-person awareness of himself (dhāt) is such a conceptualization of humanity? Surely, one may be aware of the existence of something, including oneself, without knowing what that thing is, and it is precisely such an awareness of existence that the flying man has.

Moreover, later on in Shifāʾ: Nafs (V.7, 256–57), Avicenna engages with precisely the question of whether the self I am aware of as the agent of various actions is a soul. Here the question concerns the functional definition of ‘soul’ that we have argued for, and Avicenna's answer is that as soon as that definition is learned, the answer is obvious: yes, I am an instantiation of such a principle. More important still is to note that the identification requires learning the definition of soul, an option unavailable to the flying man.

In an alternative attempt at charitability, which I have defended at some length elsewhere (Kaukua Reference Kaukua2015), Avicenna designed the flying man argument to direct our attention to our being aware of ourselves independently of any further content of experience, including any awareness of our bodies, and to apply that observation as evidence for the incorporeal existence of our dhawāt, our selves, or our essences. Admittedly, if we then take the thought experiment as a decisive proof of the incorporeality of the human essence, we must judge that Avicenna committed the masked man fallacy: he proceeded from the self's epistemic, or experiential, distinctness from its body to its metaphysical independence from the body. No physicalist worthy of the name would accept the argument, and for good reason, because so far Avicenna has given us no reason to believe that our experience is transparent even when it comes to the metaphysical ground of our own existence.

As part of their method of saving the validity of the argument, Adamson and Benevich (Reference Adamson and Benevich2018: 152) argue that Avicenna's explicit mention of the principle ‘what is affirmed is distinct from what is not affirmed’ would make no sense if he were merely making a point about distinctness in experience. Thus, they think, it must denote a real distinction between the essences of the soul and the body. I fail to see why, given that even the experiential distinction between the self and its embodiment would have been controversial enough. Anyone who has taught the flying man argument to a class of undergraduates is probably familiar with some form of the counterargument that the flying man could not possibly be aware of anything, including his self, since self-awareness is necessarily embodied. This was also a common view among the theologians (mutakallimūn) contemporaneous with Avicenna, who held that we cannot be aware of ourselves without thereby being somehow aware of our bodies (for some of the background, see Marmura Reference Marmura1986 and Vasalou Reference Vasalou2007). Thus, Avicenna's insistence on the fact that ‘what is affirmed is distinct from what is not affirmed’ serves to make a phenomenological point: if it is possible to be aware of nothing but oneself, then the self is phenomenally distinct from the body, regardless of how unusual such an awareness might be. Furthermore, it seems that Avicennian thought experiments were generally intended not to yield judgments about metaphysical possibility, but as tools for distinguishing covariant but conceptually distinct empirical features (this point requires further study, but for preliminary statements along these lines, see McGinnis [Reference McGinnis2006: 69] and Kukkonen [Reference Kukkonen2014: 456–57]).

There is another way around the fallacy, but this requires compromising the intended strength of the argument. In this interpretation, the argument can only convince interlocutors who are already prone to accept that we can know ourselves immediately, something for which Avicenna does believe he has arguments but which he knows he has not yet demonstrated. Perhaps this is why he insists that the thought experiment is intended for readers who are ‘capable of catching sight of the truth on [their] own’ and ‘do not require that [their] way of thinking is set straight’ or ‘that [they] are steered away from sophisms’ (Avicenna, Shifāʾ: Nafs I.1, 15). There is a proper demonstration for the claim that the thing that functions as a human soul is an intellect and thus incorporeal and transparent to itself (for this demonstration, which has nothing to do with self-awareness, see Avicenna, Shifāʾ: Nafs V.2, 209–16), and those who find the flying man argument fallacious, including the staunch physicalist, must be countered with the demonstration. The demonstration can also provide the ground for the validity of the flying man argument, most importantly the reason why our experience of ourselves really is transparent, but not the other way around.

The argumentative nature of the flying man is betrayed by Avicenna's characterization of it as a tanbīh, or a way of directing attention to a piece of evidence that corroborates the true view (see Marmura Reference Marmura1986). It is true that Avicenna normally describes such arguments as providing a valid framework for a proper demonstration, one just has to fill in the gaps by coming up with the missing premises. In this case, the gaps cannot be filled with anything available from the thought experiment itself, but once we know that as an intellectual being, the flying man is transparent to himself, the masked man fallacy can be avoided. We would then have a relatively straightforward application of the Barbara syllogism:

Hence, the more modest interpretation of the flying man need not be uncharitable to Avicenna nor does it necessarily commit him to the masked man fallacy. In addition, I would like to claim that it is left with one central asset, for it is capable of reconstructing the flying man as a plausible argument—not one that is entirely uncontroversial, of course, but then all the more loaded with philosophical interest.

4. Per impossibile?

Adamson and Benevich make an interesting comment about the nature of the thought experiment in passing when they say that the conceivability of the flying man makes it ‘actually possible within the causal structure of the real universe, in the good Avicennan sense that God could render him existent’ (Adamson and Benevich Reference Adamson and Benevich2018: 151). While this seems like an innocent remark—surely, an omnipotent God could come up with a person floating in the air with a specific set of meteorological conditions—it is scarcely true in the broader framework of Avicenna's metaphysics. For Avicenna, creation is a process of emanation that follows necessarily from God's essence. He explicitly denies the possibility that God could have created a different kind of universe, let alone that God could interfere with the process of the unfolding of the world's existence. Thus, the inference of metaphysical possibility from mere conceivability would require radical revisions to Avicenna's concepts of God and creation, and as far as I know, nothing suggests that he would have been willing to make them.

As a result, I believe it is more natural to approach the flying man as a per impossibile argument. It is an exercise in mere conceivability, but this is not a problem if thought experiments are understood as ways of directing attention to something that is empirically available to us, but that we either altogether miss or confuse with other things it is frequently associated with. Self-awareness is precisely such a thing: arguably, most of us have no experience of being aware of nothing but ourselves, given that in the normal circumstances, we are aware of ourselves as embodied agents and subjects of cognition, constantly immersed in our mutual engagement with the world around us. The argument is designed to show that self-awareness would remain even if these features normally associated with it were bracketed. The argument thus points to something, ourselves, the existence of which we assert without asserting the existence of any body, and this is sufficient for recognizing an existing instantiation of a real essence, the quiddity and the capacities of which we can then set out to investigate by other means.