Sunday, May 26, 2013

Avicenna’s argument from contingency, Part I


The medieval Islamic philosopher Ibn Sina or Avicenna (c. 980 - 1037) is one among that myriad of thinkers of genius unjustly neglected by contemporary philosophers.  Useful recent studies of his thought include the updated edition of Lenn Goodman’s Avicenna and Jon McGinnis’s Avicenna.  More recent still is McGinnis’s essay “The Ultimate Why Question: Avicenna on Why God is Absolutely Necessary” in John F. Wippel, ed., The Ultimate Why Question: Why Is There Anything at All Rather than Nothing Whatsoever?  Among the topics of this essay is Avicenna’s version of the argument from contingency for the existence of a divine Necessary Existent.  Let’s take a look.

The argument McGinnis discusses can be found in the Najāt, with the relevant excerpt available in the anthology Classical Arabic Philosophy, edited by McGinnis and David Reisman (at pp. 214-15).  The background to the argument is Avicenna’s view that existence, necessity, and possibility are better known to us than anything we could say in order to elucidate them.  In particular, the claim that something or other exists is more obviously correct than any argument we could give for the claim would be.  And the notions of necessity and possibility are more basic than any other notions we could appeal to in trying to define them.  (Note that he is not saying that the existence of something necessary is more obvious than any argument we could give for it; on the contrary, his aim is precisely to give an argument for it.  That something or other exists he takes to be evident; and what it would be for a thing to be necessary he takes to be evident.  But whether something necessary actually exists he does not say is evident, but requires argument.)

Nevertheless, Avicenna does think that we can say something to describe the notions of necessity and possibility, even if we cannot strictly define them.  He says that something that is “necessary in itself” is something that is entirely determinate in itself and thus requires no cause, so that if it exists it could not fail to exist under any conditions.  By contrast, something that is “possible in itself” is something that is inherently indeterminate as to its existence or non-existence, and thus requires a cause.  Again, though, these are not definitions in terms of better known or more basic concepts, but rather just criteria for identifying what would count as a possible thing or a necessary thing.  (Avicenna also identifies a third category of what is possible in itself but necessary through another.  That would be something that of itself need not exist but is nevertheless necessarily caused by some cause.) 

So, is there something that exists in a necessary way?  That brings us to Avicenna’s argument, of which McGinnis gives an exposition over several pages.  What follows is my own outline of McGinnis’s statement of the argument.  (McGinnis does not put things in this step-by-step way, so the reader should not assume that he would necessarily agree with every detail of my reconstruction.) 

Here, then is the argument:

1. Something exists.

2. Whatever exists is either possible or necessary.

3. If that something which exists is necessary, then there is a necessary existent.

4. Whatever is possible has a cause.

5. So if that something which exists is possible, then it has a cause.

Let’s pause briefly.  You might expect that after step (5), Avicenna’s strategy would be to argue that we must rule out an infinite regress of causes.  But that is not his approach.  Instead he turns his attention to the metaphysical status of the totality of possible things (where the question of whether this totality is infinitely large or not is not in view here).  Returning to the argument:

6. The totality of possible things is either necessary in itself or possible in itself.

7. The totality cannot be necessary in itself since it exists only through the existence of its members.

8. So the totality of possible things is possible in itself.

9. So the totality of possible things has a cause.

10. This cause is either internal to the totality or external to it.

11. If it is internal to the totality, then it is either necessary or possible.

12. But it cannot in that case be necessary, because the totality is comprised of possible things.

13. And it also cannot in that case be possible, since as the cause of all possible things it would in that case be its own cause, which would make it necessary and not possible after all, which is a contradiction.

14. So the cause of the totality of possible things is not internal to that totality, but external to it.

15. But if it is outside the totality of possible things, then it is necessary.

16. So there is a necessary existent.

Note that in step (13) the idea of self-causation is raised.  Avicenna does not actually think that such a thing is possible, but is merely allowing it for the sake of argument.  His point is that if a possible thing were its own cause then it would be entirely determinate in itself and rely on nothing outside it, in which case it would not really be possible but necessary.   Since this is a contradiction, what led us to it -- the assumption that the cause of the totality of possible things is internal to the totality and thus itself possible -- must be rejected.  Of course, if we simply reject the possibility of self-causation out of hand, the same result follows more quickly.

As McGinnis notes, among the distinctive features of this argument are that it not only does not require a premise to the effect that an actual infinite is impossible (as cosmological arguments often do), but also does not rely on a premise to the effect that the world of possible things is orderly (as a teleological argument does), or that it is in motion (as an Aristotelian argument from motion does), or is multiple as opposed to unified (as a Neoplatonic argument might).  Its aim is to show that if anything so much as exists at all then there must be a necessary being. 

What should we think of this argument?  And what can we know about the nature of this necessary existent?  We’ll return to these questions in another post.

43 comments:

JesseM said...

"7. The totality cannot be necessary in itself since it exists only through the existence of its members."

I don't understand what the argument for this would be. (obviously Spinoza would not agree!) A mathematical platonist would say that the infinite sequence of whole numbers exists necessarily, but isn't it true that this series exists only through the existence of its members?

Radik said...

Interesting. Bernard Bolzano makes a similiar argument in the first volume of his "Textbook of the Science of Religion". (Bolzanos logic is also highly interesting in that it anticipates many concepts and methods of Frege by several decades.)

@JesseM

I think this means something like that the existence of a set depends on the existence of its members. If A is a set of objects a, b, c, then A exist only if a, b and c exist simultaneously. If one of the members does not exist, then the set A does not exist.

Similiarly, if it is possible for one member of the set A not to exist, then it is possible for the set not to exist. Therefore, the modal status of sets (or totalities) depends completely on the modal status of their members.

Eduardo said...

JesseM

What would Spinoza say?

Now think about it this way, we have 5 necessary entities, and we call the group of all these entities ALPHA. Does Alpha could fail to exist even though none of it's members can fail to exist?

rank sophist said...

Huge thanks for writing about this topic. I find the early Islamic philosophers fascinating, and this argument is totally new to me. Can't wait for the follow-up.

John Burford said...

I'd also be interested in seeing Professor Feser discuss al-Ghazali's Kalam Cosmological Argument. The purely philosophical version, not the scientific version.

I know Professor Feser is aware of it (especially through the work of William Lane Craig), but I haven't seen any articles from him on it.

I know a lot of Thomists, including Aquinas himself, reject it, but Craig's argument against the possibility of an actual infinite in his book "On Guard" seems convincing to me.

Jinzang said...

If there's a flaw in the argument, here's it: the totality of all possible things is not a thing, so the premise that all things are either possible or necessary does not apply to it. And without this, the rest of the argument fails.

Scott said...

@Jinzang:

"If there's a flaw in the argument, here's it: the totality of all possible things is not a thing, so the premise that all things are either possible or necessary does not apply to it. And without this, the rest of the argument fails."

The premise is not that "all things are either possible or necessary" [emphasis mine]; the premise is that "[w]hatever exists is either possible or necessary" [emphasis again mine]. The totality of possible things does exist, does it not? Or does it?

rank sophist said...

Jinzang,

I assumed that the "totality of possible things" was a set that referred to individual entities 1 through ∞. The totality itself is just an abstraction based on the particular instances of possible being. I may be wrong, though.

dd said...

rank,

see the following:

http://kimiyagard.wordpress.com/2011/07/11/on-the-infinite-regress-assumption/

http://kimiyagard.wordpress.com/2012/03/27/from-the-contingency-of-essences-to-the-existence-of-the-necessary/

http://kimiyagard.wordpress.com/2013/01/17/ibn-sina-contra-al-ghazali-and-h-davidson/

Douglas Ryan said...

Having difficulty with premise 4. I can think of two readings:

4a Every existent possible being has a cause.

and

4b Every possibly existent being has a cause.

4a looks equivalent to the familiarly disputed claim that everything that exists contingently has a cause, since something exists contingently iff it is possible (in Avicenna's unusual sense) and exists.

4b looks too strong: after all, possible beings that lack existence lack causes too.

Tony said...

Scott, it seems to me that "the totality of all possible things" cannot be said "to exist" in any simple sense. My dog is "a possible thing" in some valid sense - she existed, and she wasn't a necessary thing when she existed. But she died. Since she does not exist now, she is not now a contributory member of any set that exists now, for which we could say the set "exists" in dependence on its members existing.

If we want to be clearer, her existence is conceivable, we can conceive of her existing, and one may formulate "sets of conceivable things." The existence of a set of conceivable things doesn't, however, depend on the conceivable things existing. It depends, rather, on the possibilities of the mind, not of the beings - it is a rational "being" not a substantial entity.

Which also pertains to the above-mentioned set of all whole numbers. Assume for the moment that the physical universe is finite, holding a finite number of particles, of quanta of energy, etc. There is a finite number that is an upper limit to the number of all possible relations between these things. There is, then, no "thing" that any higher number could actually name in relation to all other things. It would, perforce, be a name only of mind, in a less real sense: a rational relation only. Does such a number "exist"? What does it mean for a number to exist?

kuartus said...

The argument seems similar to Robert C. Koons' new cosmological argument.

Scott said...

@Douglas Ryan:

"4a looks equivalent to the familiarly disputed claim that everything that exists contingently has a cause, since something exists contingently iff it is possible (in Avicenna's unusual sense) and exists."

That appears to be what's intended, as what follows is this: "So if that something which exists is possible, then it has a cause" [my emphasis].

@Tony:

You've put your finger on my questions/misgivings about whether possibilities can be said to "exist." The modern neo-Platonist John Leslie, for example, would (if I read him correctly) say that they do not, but that they are not therefore unreal; for him, "there is something rather than nothing" because there just plain can't be a "nothing" that doesn't even include a Platonic realm of possibilities, and given that realm, some possibilities are actualized (exist) because they are in some important way "required." In his view as I understand it, "being real" and "existing" mean two different things, and possibilities are in some way "real" without (necessarily) "existing" or being actualized.

Discussing that in detail would probably take us pretty far afield. But for the present discussion, it's probably sufficient to observe that on a neo-Platonist view like his, Avicenna's argument might not go through, as it may not make sense to describe "the totality of all possibilities" as existing either necessarily or contingently (in the A-T sense of those terms), because in fact it may not exist at all.

dd said...

Tony and Scott,

by the totality of possibles, Avicenna means, not possibilities in the abstract, but possible existents in the concrete i.e., essences which, although actually existent, considered in themselves are indifferent to either existence or non-existence.

rank sophist said...

dd,

Thanks for the links. In particular I found al-Ghazali's objection interesting, since it kind of reminds me of Heidegger's immanent ontology. It's a bit like the hermeneutic circle--and it ultimately suffers from the same "concealed [...] self-existentiation" that the third link mentions. I'm excited to see the rest of Prof. Feser's take on this argument.

Anonymous said...

7 seems to be the most controversial. Whether or not one thinks that the totality of the parts of the universe is contingent depends on whether or not one thinks the universe is contingent. Although I disagree with the man who holds the material universe to be necessary, if he does so he certainly has no problem agreeing that all of the universe's parts follow necessarily from the universe's existence as well. Of course this position is absurd, because it would mean that we exist necessarily (as do our thoughts and actions). But absurdity alone won't convince the anti-realist committed to the idea that the world just truly is absurd.

So I think an Aristotilian (or kalam) Cosmological argument is better since it attacks the absurdity of a necessary, self-contained universe instead of merely assuming it away.

--GW

Joe K. said...

Tony,

I'm sorry your dog died. That's the worst.

Scott said...

@dd:

"by the totality of possibles, Avicenna means, not possibilities in the abstract, but possible existents in the concrete i.e., essences which, although actually existent, considered in themselves are indifferent to either existence or non-existence."

Ah, of course -- so by "the totality of possible things" he means what Aristotle would mean by it: the totality of all things which exist and are possible (i.e., contingent rather than necessary). He's not including things that might exist but don't; he's including only things that do exist but might not have.

In that case, I think the argument does get us at least one necessary existent but doesn't appear to guarantee (so far) that there's just one. I expect that's coming in Round Two.

dd said...

Scott,

"In that case, I think the argument does get us at least one necessary existent but doesn't appear to guarantee (so far) that there's just one. I expect that's coming in Round Two"

well, that's a distinct issue and he has demonstrations for that too. see Bk.1.5 and Bk.8.5 of the Metaphysics of the Healing (trans. by M. Marmura here: http://www.amazon.com/Metaphysics-Healing-Brigham-Young-University/dp/0934893772/ref=sr_1_1?ie=UTF8&qid=1369661972&sr=8-1&keywords=the+metaphysics+of+the+healing)

dd said...

scott,

on the topic of God's unity, you may find this helpful:

http://kimiyagard.wordpress.com/2012/08/03/ibn-sina-on-the-oneness-of-the-necessary/

Scott said...

@dd:

As I said, I'm sure Feser will address this in his next post, but thanks for the links in the meantime.

Charles said...

"As McGinnis notes, among the distinctive features of this argument are that it not only does not require a premise to the effect that an actual infinite is impossible (as cosmological arguments often do)"

Dear Dr. Feser,
I have a quibble with the above excerpt. Maybe this is a disagreement with McGinnis and not with you. My problem with the above is this: As I understand the doctrine of the Posterior Analytics and the Physics, the impossibility of an actual infinite is the conclusion of a demonstration in the philosophy of nature that precedes a demonstration of the first mover, but does not directly enter into it. The claim of St. Thomas in the first way that "non est procedere in infinitum" is rather a logical principle, and not a premise in the proof, that instructs us to find the first essential cause (the per se primo cause) of some given effect. It is in the major of each of the proofs that this cause is identified and predicated of its proper effect. So just as there is no infinite regress in discovering the source of the properties of various kinds of triangles by identifying the commensurate universal to which they belong (thus discovering the per se primo cause of these properties), so in the proofs for God the effects we use must be traced back to their proper cause, yielding a major premise that predicates the cause to the effect in the 4th mode of per se predication and thus yielding a conclusion that this proper cause exists (without, of course, telling us the essence of the cause). At any rate, it seems to me that treating the impossibility of an infinite regress as a premise yields a much more complicated syllogism than is necessary. The older Thomists generally claim that each of the five ways is composed essentially of two demonstrations: one that identifies the proper cause of a given effect, and one that identifies that proper cause with what we mean by the name God, both of which are non-convertible quia demonstrations.

Love your blog.

Joe K. said...

Slightly on-topic: does anyone know how to pronounce "Avicenna?" I have three candidates, but they could all be wrong:

a-vi-CHEN-uh (like Italian or some Ecclesiastical Latin, though this seems wrong to me)

a-vi-SEE-nuh

a-vi-SIN-uh (on the New Advent page, the Latinization apparently comes from "IBN SINA," but I have no idea how to pronounce "sina" in Arabic or whatever it is)

I'm putting the accent on the penultimate, which is why I want to pronounce it like the Italian, I think, but again, I really don't know.

If I were talking to a Western philosopher about him, though, what would be considered correct? I kind of want to talk to a friend about him, but it's annoying not knowing how to pronounce a name.

MarcAnthony said...

This is really, really interesting. It looks to me like it's almost a scholastic approach looked at from a completely different angle. Really looking forward to your next post on this topic, because it's fascinating. I'll have to learn more about Avicenna.

Dianelos Georgoudis said...

I’d object to #4 “whatever is possible has a cause”. I see no logical impossibility in X existing, not existing necessarily (i.e. not forming part of all possible worlds in the relevant sense of “possible worlds”), and not being caused by anything - but just simply being in some worlds but not in others.

I have a larger problem. A common ambiguity with modal arguments is that they don’t specify their domain of possibility. The domain of logically possible worlds is clear enough: By “logically possible worlds” one means the worlds whose book (i.e. the set of all true propositions) is logically consistent (i.e. devoid of logical contradictions). But here it is easy to prove that there is no necessary existent, since the empty world where nothing exists is logically possible (its book is empty and thus devoid of logical contradictions). Therefore there is no existent which is present in all logically possible worlds.

So any argument for the claim that there is a necessary existent must define a narrower set than the set of all logically possible worlds. One often speaks of “metaphysical possibility”, defined in such a way that, for example, the empty world is not metaphysically possible. One useful definition of the domain is to specify that any possible state of the actual world is a metaphysically possible world. (And for simplicity’s sake define that the actual world is what we experience plus what it is reasonable to assume must exist in order to metaphysically ground it.) But now the claim “there is a necessary existent” becomes trivial, for a necessary existent is the actual world itself, since by definition it is present in all elements of the set.

My argument in short is this: Any modal argument with the conclusion “there is a necessary existent” either does not define the domain (the set of possible worlds), in which case it is ambiguous and fails, or else defines the domain, in which case which existents are necessary or possible is only a function of *how* the domain is defined. But then such an argument cannot really teach us anything about reality.

Matthew Petersen said...

The with that argument is that it assumes a particular understanding of possibility, which offers a relatively good approximation of possibility and necessity in the limited sense that existing things can be necessary or unnecessary. But Avicenna is asking about the whole of things that exist, including the collection of sets of possible worlds, and so it makes no sense to appeal to that collection, or to any of the particular sets in that collection, in his definition of necessity. To do so is to define necessity in terms of something contingent, so that if we say X is necessary, we mean X contingently has property Y--that is, we say nothing about X.

JTH said...

Does anyone here know of a good discussion of ab alio necessity, especially in Averroes?

machinephilosophy said...

Thanks, Ed. This argument is new to me as well, and I think it successfully bypasses infinite series issues.

Also, information-theoretic considerations seem to confirm this. The argument could be expressed in terms of object-oriented analysis, design, and programming, which considers any set of objects to itself be an object, which is the key to the advantages and success of object-oriented development.

Also, any attempt to deny existence to either the whats ("whatever") or the aggregate, runs up against the fact that existence is already embedded in the meaning of the verb to be, an issue existence deniers seem to always avoid in their own denials. Surprise, surprise.

Pedro Erik said...

Dear Feser,

I am loving your book (The Last Superstition). The best book to explain the sources of the evil today. I recommended it in my blog to Portuguese readers.

But, as you know, in Brazil or Portugal not many people can read in English.

Are not interested in translating into Portuguese?

Contact me, if you are interested.

Best regards,
Pedro

Michael Brazier said...

The step from 9 to 10 is invalid. Up to 9 "the totality of possible things" means the class of whatever is contingent, considered as a unit. In later steps, however, the cause of the totality of possible things has to mean not the class, but the members of the class, for the argument to hold. Specifically, in 13 the cause of the totality of possible things can be a member of the class without causing itself. Say I found a business partnership; the class of partners in the firm exists because of me, yet no member of the class exists because of me. And I do not become self-existent just because I am one of the partners, and thus a member of the class that I brought into existence.

Put differently, the step from 9 to 10 is a quantifier shift fallacy, going from "for everything in the totality, there is some cause" to "there is something that is the cause of everything in the totality."

Asadullah Ali said...

Salaam Professor,

Thank you for drawing attention to one of our great thinkers. May I ask if you've read up on the contemporary Islamic philosopher, Seyed Muhammad Naqib al-Attas?

Anonymous said...

Michael Brazier,

If the totality of possible things is nothing other than all possible things, then in order to cause the totality, you have to cause each member. Your example doesn't work, because a partnership is not merely the members of the partnership, for the members existed prior to the partnership. Rather, the partnership is a particular relation of members, and it could be the case that a member caused that relation. On the other hand, it cannot be the case that a member of possible things caused the existence of the unit of possible things, for this is nothing other than causing the existence of all possible things, including its own existence, which is impossible.

Wallahu A'lam

Anonymous said...

Dr Feser,

This is Peter Adamson's exposition of Ibn Sina's argument.

http://ec.libsyn.com/p/a/e/0/ae0c36acd9cbfef7/AdamsonMixSes.MP3?d13a76d516d9dec20c3d276ce028ed5089ab1ce3dae902ea1d01ce8e3fd4cb59a85f&c_id=1779083

Tap said...

Would be interested in Dguller's take on this since he has an arabic background

Robert said...

6. The totality of possible things is either necessary in itself or possible in itself.

Though totality can be a null set and still represent the totality. I suppose that the concept of totality is necessary by definition, but it seems like an odd way to use this word in relation to this particular argument.

Tony said...

I’d object to #4 “whatever is possible has a cause”. I see no logical impossibility in X existing, not existing necessarily (i.e. not forming part of all possible worlds in the relevant sense of “possible worlds”), and not being caused by anything - but just simply being in some worlds but not in others.

Danielos, dd answered this above. In Avicenna's argument, "possible things" refers to things that actually exist right now, at the moment you are making the argument, that are "possible" rather than necessary. That is, the "set" is the set of now existing things that are not necessary: they may exist but they are essentially indeterminate as to existence or non-existence.

The step from 9 to 10 is invalid. Up to 9 "the totality of possible things" means the class of whatever is contingent, considered as a unit.

Michael Brazier, I think your issue is really with #6-8, in treating "the totality" as "a thing", an existent.

It would be theoretically possible to say that "necessary" and "possible" are categories one of which must inhere in (characterize "things" that are real things, substances for example) but need not belong to analogous "things" that are not substances.

Suppose I write a fable in which the divine source of all consists of 2 beings "the all father, Odin" and "the winged unicorn, Jewel". In the story, Odin and Jewel are characterized as being necessary. Properly speaking, nobody outside of the story is forced to say whether Odin and Jewel are necessary or possible - the sense in which they "exist" does not require that they be one or the other. Likewise, perhaps the "set of all possibles" is not the sort of "thing" (i.e. is not a substance) of which one must be able to posit that it is either necessary or possible. Perhaps, since it is "existing" only in a sense, the characterization "necessary or possible" does not apply to it.

Eduardo said...

dguller is ... arabic descendent XD????

dguller said...

Tap & Eduardo:

Sorry, I can't comment on Ibn Sina, because I haven't studied him much. The most that I've read is Gilson's analysis of his essentialism in Being and Some Philosophers. And no, I'm not Arabic, but I was a practicing Muslim for many years before I lost my faith. As a traditional Muslim, I was more interested in Imam al-Ghazali, whose Incoherence of the Philosophers was a sustained refutation against Ibn Sina. After my reading of Aquinas, I think I'll have to give Ibn Sina another look. :)

Eduardo said...

lol D, that explains why you are not as angry as the ex Christian Atheys!

U_U HA! I knew there was a perfectly logical explanation!

יאיר רזק said...

I would appreciate help in understanding the concept of "possibility" and "necessity" being employed here.

As I read the argument, the intuition at work here is contingent-on (X). Something is said to be "possible" if it is contingent-on-something-else. Something is said to be "necessary" if it is contingent-on-itself. It is implicitly (?) (a) denied that something can be contingent-on-nothing (i.e. not contingent-on anything), and (b) asserted that this dichotomy applies to sets of things as well as to individual things. This principle is applied to things that actually exist - every existing thing, and every set of existing things, is contingent-on itself or on something else.

Where I am confused is on the relation between this concept and the concept of could-have-been-otherwise contingency, which is captured by speech of possible worlds. I'll call this logical-contingency. Something is logically contingent if there is a possible world where it holds, i.e. if there is no logical contradiction in it, and something is logically necessary if it holds in all possible worlds, i.e. if it is a tautology that holds always.

It appears to me that the two concepts are distinct. But then again, it also appears to me that the Scholastic/Aristotelian view probably identifies the two. So I'd appreciate it if someone could illuminate why and how contingent-on and logical contingency are connected from a Thomist / Aristotelian / Avicenna-ic viewpoint.

Thanks,

Yair

Kjetil Kringlebotten said...

@Douglas Ryan, 4b looks too strong: after all, possible beings that lack existence lack causes too.
@Scott [to Tony, not Ryan], You've put your finger on my questions/misgivings about whether possibilities can be said to "exist."

Yes, but that doesn’t seem to be the way Avicenna uses the term. Premise 2 reads: “Whatever exists is either possible or necessary.” It seems that possible is here used about things that do exist, but only possibly, i.e. that it is possible for them no to exist. I don’t think he uses the term about non-existent things that could possibly exist, like possible universes, unicorns, yourself in a different height or skincolor, etc. So it seems that what Avicenna means by ‘possible things’ or ‘possible being’ is ‘contingent things’ or ‘contingent beings.’

Anonymous said...

So when's part 2?

Edward Feser said...

So when's part 2?

Actually, I've just been finishing writing it up this evening. It will be posted shortly.