Saturday, August 11, 2012

The Münchhausen Trilemma, the First Cause and the Infinite Regress

This is yet another topic that I have put off for far to long. Initially, I wanted to make a youtube video on the topic, but it never came to be, so here it comes in written form, which is probably better anyway. It's going to be a long ride, please bear with me.

The Starting Point

One favourite argument of First Cause proponents is a reductio ad absurdum of the opposite. I get to hear it from theists, but of course you don't have to be a theist to put it forward (except in the strict philosophical sense, where god is simply the absolute, not necessarily a person or an intelligence).

The argument works something like this:
  1. Every event has a cause.
  2. An infinite regress of causes leads to an absurd result.
  3. Therefore, there had to have been a First Cause.

The Reductio In General

A reductio ad absurdum is, of course, a valid form of argument - some go so far as to say that it's the queen of arguments. It works to exclude answers from the spectrum of possibilities. If an answer leads to impossible results, then that answer cannot be true. That's just simple logic.

This is not at all far-fetched or highly philosophical. Proofs of that nature occur, for example, in computer science, where it is customary to show that if a proposition were true, then it would solve the halting problem. Alan Turing proved that the halting problem is unsolvable, therefore the proposition in question is false. (If you suspect that I have a weird tendency to bring Turing into my articles, you're right. Read up on his biography - the guy was a genius, a war hero, an uber-geek, and a highly tragic figure. He committed suicide by eating a poisoned apple. No kidding.)

It's a way to rule out impossibilities - the Sherlock Holmes way of excluding impossible options so you can focus on those that are indeed realistic.

If you try to use this argument to prove one specific possibility, your success hinges on the following conditions:
  1. The premises of your argument must be agreed upon.
  2. Your argument must be consistent.
  3. You must show that you have actually excluded all options except the one you're trying to prove.
  4. The one option you're proposing must, itself, not lead to absurd results.
So let's have a look at the reductio argument for a First Cause to see whether it meets those criteria.

The Premises

"Every Event Has a Cause"

Can we agree that every event is caused by something else? At first glance, it certainly seems that way, as we would be hard-pressed to come up with an event of which we can show that it is uncaused. (As always in such discussions, I hereby make a sacred vow not to invoke quantum theory.) We know that everything is caused - we know it from our experience. Indeed, we know it from every experience that any human being ever had. That's an awful lot of experience!

Aye - but it's still experience. And that's a bummer.

It's called inductive reasoning. A synthetic proposition a posteriori, as Kant would have called it. What that means is that, while it seems extraordinarily probable that the next event we encounter will have a cause, just like all the events we encountered in the past, we don't know that for sure. We cannot be absolutely certain. We can predict, with almost 100% certainty, that it will be the case - but almost 100% is not the same thing as exactly 100%!

So, actually, no, we cannot agree on the first premise. Not to the degree that would be necessary for a point about the absolute, a priori, First Cause.

(Oh, and, by the way - if the First Cause caused everything else, where the heck did causality come from? If the First Cause caused it, then using it to prove the First Cause is, itself, circular. If the First Cause didn't cause it, then what caused it, and how can the First Cause be said to be... well... the First Cause?)

"An Infinite Regress Leads to an Absurd Result"

This is, obviously, the core of the argument. The logic here is pretty impressive and has fascinated me from my first adolescent forays into philosophy until now:
  1. There is an infinite chain of cause and effect.
  2. An infinite chain cannot be traversed.
  3. In order to ultimately explain any given event (like, this moment right now), you have to traverse it's causal chain.
  4. Therefore, you can never explain any event.
This is impeccable logic. For as long as I thought that ultimate explanations are a good and even necessary thing, it caused me lots of pain. Then I started to relax. Now I find it rather amusing.

Of course, on top of relying on causality being universal and absolute, this introduces even more premises that have to be agreed upon.

Sidenote #1: Is Causality a Property of Things?

It's an interesting question to ask yourself: What exactly IS a cause? When we say that B is caused by A, do we actually mean that B has an intrinsic property that can be called "is caused by A"? Is a cause a property of things?

I'd like to introduce a little example here. Picture a child throwing a stone at a window. Predictably, the window breaks. But what exactly is the cause of the window's demise?
  • The child is the cause? That's one intuitively correct answer, of course.
  • The stone? Yet another correct answer.
  • The momentum that the stone put into the glass?
  • The child's intention in throwing the stone?
  • The child's brain's activity that represents said intention? (Note: it does not CAUSE the intention. It IS the intention, on a biological level.)
  • The bad upbringing of said child by her parents?
  • Society at large?
I'm sure that you can come up with a few more examples, some more useful than others. Some might even be comically absurd or mere esoteric hypotheses.

(And of course, you can put most of them into an order of temporary succession. Did I mention that causality is not necessarily temporal? I won't go into it here, it's a fun little riddle for long lonesome (k)nights.)

The point is, though, that you can invent any number of causes. Some are practical and some might be rather obscure, but they're still possible answers. Nobody can keep you from naming the air that the child breathed as one cause of the child's throwing the stone.

This may lead us to the hypothesis that causality is not simply an attribute of things. If it were, then how come we can come up with any number of possible answers to the simple question "What caused event X", all of which are equally true? If causality were a naturally existing, universal attribute of things outside our brains, then there should be exactly one answer to this question, and everyone should be able to agree on it (stupidity and willful ignorance aside).

I propose that causality is, at least partially, a result of human thought processes. Of course, our brains evolved to work in the real world, so most probably there is some kind of correlation to natural phenomena outside ourselves. But it is not fair game to simply call this outside factor "causality" and leave it at that.

Sidenote #2: Is Causality a String of Events?

In order for the original argument to work, causality must be thought of as a string of causation leading to the past. Picture it as a string of pearls put on a chain. A very long chain. Very, very long.

I think that this is an oversimplification. It is useful (and customary) to distinguish two different classes of causes: necessary and sufficient causes.

"The street is wet." - Well, it rained the whole day. That would be sufficient to explain the wet street. Since there were no other events to explain the effect, it is also its necessary cause.

On the other hand, picture a slightly more complex scenario: A wedding. It is highly recommended for the bride to be present. The same goes for the bridegroom, the priest, and the witnesses. On top of that, it's rather practical if they all agree to have a wedding, and if nobody shows up at the last moment to speak up now, rather than be silent forever. So we can easily say that every person's presence is a necessary cause (all of them, seperately), while the whole array of conditions together form the sufficient cause.

Ah-ha! It seems that there may be any number of necessary causes, and all of them together can be said to be the sufficient cause of an event.

So, if we look only at sufficient causes, then we do get our string back... or do we?

Remember the child from further up. The child had to throw the stone a certain way (necesary). The glass had to be there (necessary). The wind had to blow the right way and strength (necessary). Together, they form a sufficient cause. But every single one of those necessary causes depends on a new string of causation. It is not easy to see how this tree of causation would, at some point, magically collapse into one simple string that leads to one First Cause.

If causality were just a succession of pearls on a chain - shouldn't it then be trivially simple to point out the one cause that immediately precedes any given effect?

I'm not saying it's impossible. But it's yet another telltale sign that the idea of a string of causation leading back to the First Cause is just a bit too simplistic.

Sidenote #3: Do We Have Infinite Regress Anyway?

Furthermore, it seems to me that we can make up any number of causes in-between causes.

"The child threw the stone, and it caused the glass to break." - Yes, and in between this and that, the stone followed a certain ballistic curve. At about 1/3rd of its way, the wind moved the stone slightly to the left. So the position of the stone at this point, its momentum, and the wind, ultimately caused the glass to break. Yes, and about another 1/3rd from this point to the window, a bird flew by and caused pressure in the air, which moved the stone again. So ultimately, the bird caused the breaking of the glass.

See what I'm aiming at? At every single position along its way, it can be said that the stone's current position, plus its momentum, caused its further path.

Between any event and its direct necessary cause, you can always find another cause. Hypothetically, this goes on ad infinitum (again, not taking quantum mechanics into account).

It's a bit like natural and real numbers: Between any two natural numbers, there are always infinitely many real numbers. You cannot traverse them. You can only jump over them to reach the next natural number.

But if that is the case, then it doesn't matter whether we have a First Cause at the beginning. Right between the First Cause and its First Effect, there was already an infinity of causal relationships. We have an infinite regress anyway, always, regardless. So the First Cause doesn't actually solve the problem. Instead, it introduces one more problem by setting up a dogma for no good reason.

Did We Cover All Options?

As stated above, if we want to use our argument from absurdity to show that our favoured option is true, we have to make sure that we excluded all possible options.

The Third Option: Circular Logic

As the title of this article suggests, there is at least one additional option that we have not covered.

In its classical form, the Münchhausen trilemma states that, on top of infinite regress or a first cause, you can always resort to circular causality. Of course, as we all know, circularity is a logical no-go. So is the infinite regress, and so is dogma. That's the whole point of the trilemma: All options are equally absurd.

But I'm aiming at something else here. I'm not exactly certain whether the third option actually is one. Neither am I convinced that it's not. And, reading a few articles about it, I get the impression that really, nobody is to certain in that regard. Circularity might be seen as a variation of the infinite regress, but there is also a point to be made that it's a variation of setting up an absolute. Or that it's a third, independent option. I don't know. And as long as there is no good argument to exclude any of those options, we cannot exclude the possibility that there is at least one more possible answer to the question of the ultimate cause. Maybe there are even more? How would you go about showing that such a thing is impossible?

Ironically, the official doctrine of the catholic church, which obviously opts against infinite regress and in favour of an absolute First Cause, is suspiciously reminiscent of the circularity variation: All existing things need sustenance, but god provides its own sustenance and is therefore absolute... or... circular... or... infinitely regressing... or... To me, this has always seemed a lot like an excuse. They don't want to put the absolute First Cause right in your face, but they want to set it up anyway, so they invent something like an "additional realm of being", in which circularity is somehow allowed.

So, Do I Propose An Infinite Regress?

Short answer: no. I cannot stress this enough. My dear friends on the theist side always seem to get this wrong, pretty much regardless of what I say. I am not proposing that an infinite regress is a good solution. I concur that an infinite regress leads to an absurd conclusion. But I also state that setting up any one "thing" (i.e., god) as an absolute First Cause cannot save us from the anxious question of what caused it. And of course, circularity doesn't help us either.

I think it's an unsolvable problem. And I think that there is a very good reason for why it is unsolvable.

If Causation Is Not A Property Of Things, What Is It?

I think the answer lies in the fact that causation is not a property of things. We should treat it more like part of our model of the universe, rather than like part of the universe itself. A highly successful part of this model, mind you - but still, the model is not the airplane.

That way, we can opt for either one of the three options, depending on what is more practical in a given context.

For example, with regard to societies and human interactions, a somewhat circular approach is probably more than appropriate: Complex systems of interactions looping back to their causes.

With regard to classical Newtonian physics, of course, linear causality makes perfect sense: the ballistic curve is strictly deterministic.

I don't have an example for where it's useful to use a dogmatic First Cause. I'm hesitant to say it, but it might actually make sense in cosmology. It's perfectly okay to say that your model of the whole universe is based upon a First Cause, simply because it may be rather practical to do so - as long as you don't think that you're actually talking about actual properties of the universe itself, there is no issue with that.

Of course, there is probably a very good reason for why we don't seem to find the answer to Münchhausen's trilemma: Causality may be hardwired right there in our brain. The reason for that being that it just works great within nature. (Now we're close to circularity.) The reason for that being that there is a correlate of causality in the universe, which itself has no reason at all. (A dogma!) Or which is caused by something else which we haven't discovered yet, which is caused by something else which is caused...

Summing It Up

So... after all this long-winded talking, here's the executive summary of my contentions:

  1. Causality is only derived from experience.
  2. Causality is not necessarily linear.
  3. There may be infinite causes between each pair of cause and effect.
  4. Causality is not a property of the universe, but rather of the way our mind works.
  5. All answers that we can give are equally unsatisfying.
  6. There may not be an ultimate answer.
If you enjoyed this, you might want to take a look at the addendum here.


  1. brent.mosher@yahoo.comAugust 13, 2012 at 2:32 AM

    Thoughtful essay. Thanks for writing it...I like the way you think.

    1. Thanks. It took a very long time to find out where I'm standing at with regard to this, especially when it comes to the finer points. Basically, I've been thinking about it for twenty years, again and again. ;-)

    2. ReThinkingTheism here. Thanks for commenting on my videos.

      I try to leave this issue at epistemology. That is, no one has epistemological access to events prior to the big bang (or creation if that's the route you want to take).

      The argument I've been interested in lately is the possibility of the big bang being caused by (meta-)natural causes. How can we tell if it was meta-natural or designed by a benevolent entity? The bottom line is that we can't. For me, that's the end of the discussion.

      It seems to me that infinite regress is a product of the human mind that arises when it applies logic to things that it really shouldn't. Just saying, "I have no freakin' idea where this all came from" shatters the dependence on God (that upon which all things are absolutely dependent) -and- infinite regress.

      Both of these are asserted as logical necessities to try to patch up holes in our stories. Who said our story needed to be complete anyway???

    3. Ultimately, I guess, our stories are never complete. :-) On a personal note, I think I have some idea where you're coming from. I tried to adhere to the agnostic stance with regard to metaphysics for a while.

      Interestingly, that was at a time when I had a fling with buddhism and advaita vedanta, much as (it seems to me, at least) you're doing now.

      Well, in a way, in this article, that's what I'm still doing. After all, I'm saying that all possible answers lead to impossible implications.

      Sure, refraining from any answer is one possible way to deal with it. In a certain manner, it's probably the "cleanest" and most honest way. Me, I can't help but attack the problem again and again. Attachment, you might say; though, as far as attachments go, this is definitely not the worst I can think of. ;-)

      I just can't stop being fascinated by this issue.

      Probably, to some degree it's a matter of personality, but seeing how many folks seek out some form of religion, I have the impression that it's a part of the conditio humana that we're incapable of not filling in the holes with some kind of answer.

      Out of curiosity, does Ockham's Razor not convince you that the default position should be that the big bang is not designed by a benevolent entity (because it's an additional entity)? It would be interesting to find a counterargument to Mr Ockham, as I've never managed to find one.