Is there a cogent argument against the principle of sufficient reason?



As far as I can see, there are no significant arguments against the principle that all events have a cause, which is to say the principle of sufficient reason. (It's important to note that the seemingly identical idea that all effects have causes is a circular argument based on the mutual definitions of "cause" and "effect".) While the idea seems intuitively obvious and therefore self-evident, we hold many counter-intuitive ideas to be true.

Has anyone proposed a serious argument that events sometimes are not caused?

Clarification: The question title may be misleading because it suggests that the question is an epistemological one, but my actual question is metaphysical (or perhaps even ontological). Whether or not we can always (or even ever) know the sufficient reasons for an event is beside the point (unless it can be shown that we always can know the cause of every event).

I've been asked to define what I mean by an event. That's a bit more than I can take on at the moment, but the Stanford Encyclopedia of Philosophy suggests that we have a "prima facie commitment to entities of this sort." If I had to suggest a definition, I'd say an event is a discrete observation or inference about a period of time. That I was married is an event that was observed by many people. That the sun was formed is an event inferred by the current state of the universe. Of course, that definition has an assumption buried in it that makes the question less interesting: inference implies causation. So we need to find a definition that conforms to our intuition of what an event is, but does not implicitly conform to our intuition that events are caused.

For the purposes of this question, the best definition of an event is that it is something that happens. Do things happen for which there is no cause?

Jon Ericson

Posted 2011-11-17T06:25:09.187

Reputation: 6 843

1+1 I've always dismissed the idea of asking this question myself because I thought the answer to be blatantly obvious (no). However, it is always better to ask and get "no" for an answer then not ask at all and never know for sure. :) I eagerly await references to literature on this idea (if they exist)... ^_^ – stoicfury – 2011-11-17T07:05:46.243



Actually, there are a number of significant arguments against the principle of sufficient reason; you can find them in Sextus Empiricus, Hume, Wittgenstein and Nāgārjuna to name but a few.

In terms of accessibility, I suppose I'd recommend starting with Hume's view, which you can read about here or here, followed by Wittgenstein on rule-following, which you can read about here.

Of course, if you are more familiar with classical literature, you can check out Sextus Empiricus (Outlines of Pyrrhonism, Book III, if I recall correctly); similarly, if you are more familiar with Buddhist philosophy, you can begin with Book I of the Mūlamadhyamakakārikā.


I should clarify that none of these thinkers suggest that there are events that are uncaused; this is one of the positions that Nāgārjuna explicitly rejects in the first verse of the Mūlamadhyamakakārikā. Rather, each calls into question the notion of causality, and attacks either the notion of "sufficiency" or the notion of "reason" with regard to the matter.

Michael Dorfman

Posted 2011-11-17T06:25:09.187

Reputation: 22 863


Although I haven't read Sextus Empiricus, Nāgārjuna, and (regrettably) Wittgenstein at any length, I know that although Hume found no "necessary connexion" between events, he didn't actually suggest that there are events which are uncaused and he agrees that we are still forced to live by live by our notion of causality: "...We are nonetheless always determined to proceed in accordance with this supposition. There is a natural basis or “principle” for all our arguments from experience, even if there is no ultimate foundation in reasoning (EHU 5.4–5; SBN 42–43)".

– stoicfury – 2011-11-17T16:57:22.330

2@stoicfury: Good point. I was attempting to answer the question in the title, not the final query in the text of the question. I'll edit later tonight to reflect this point. Thanks. – Michael Dorfman – 2011-11-17T17:04:13.717

So this is really an answer to the question of "can we know the cause of all events?" not "do all events have a cause?" The first is really an epistemological question whereas my question is a metaphysical one. ;-) – Jon Ericson – 2011-11-17T21:36:20.660

@JonEricson: It's an attempt to answer "Is there a cogent argument against the principle of sufficient reason?" which has metaphysical and epistemological components. One easy entrance to this is through contemplating the logical fallacy post hoc, ergo propter hoc. – Michael Dorfman – 2011-11-18T07:49:34.017

I updated my question. Do you know of another (less equivocal) term for what I'm asking about that would fit in the title of the question? – Jon Ericson – 2011-11-18T20:11:26.303

I don't see how the fallacy addresses the metaphysical question at all. Surely it is fallacious because if we pick one event (or set of event) that came before and anointed it as the cause of an event that comes after we risk picking the wrong cause. Analysis may require us to say the cause of an event is the state of the entire universe prior to the event, which is unwieldy but still sufficient cause. – Jon Ericson – 2011-11-18T20:18:23.823

2The question comes down to what we mean by "causality", which is much more complex than people realize. What is the difference between saying "A happened, and then B happened" and "B happened because A happened"? We might suppose, inferentially, that if B happens every time A happens, then we can say that A causes B; but here we are into issues of Humean regularities and Wittgensteinian rule-following. So let us switch tacks for a moment, and suppose there were an uncaused event-- how would we know? What would be the distinguishing characteristic of said event? – Michael Dorfman – 2011-11-18T20:34:12.513

Ah. I see where you are headed now. Of course one would search for a cause to each unexplained event and if sufficient cause were not found, the principle of sufficient causes could be appealed to--a cause must exist. But that would be circular. Therefore, such a belief can not be shown to be true. On the other hand, given the choice between the principle and rejecting the principle, that we can't prove it seems a small thing compared to the fact we all behave as if it were true. Pushed far enough for certainty, I can't prove anything at all. – Jon Ericson – 2011-11-18T21:28:11.423


I'm not sure the idea of "reason" is sufficiently specifiable for the question to make sense. I am not familiar with arguments that events are not caused, but there are at least a couple of reasons to be worried.

Since quantum mechanics seems non-deterministic, one could argue that things do indeed happen without reason...or one could broaden the definition of "things" and "reason" so that QM fits nicely within the box.

Also, we don't have direct access to causes; all we have is sense data about what is happening. A cause is thus a generalization of a statistical measurement on sense data; a reason invokes the appropriate conditions and causes. However, in certain cases we have dreadfully little statistical data (e.g. how many universe-creation-events have we witnessed, or does it even make sense to think of this as an event?), so there is dramatically less reason to think that all events are caused in such situations.

Rex Kerr

Posted 2011-11-17T06:25:09.187

Reputation: 15 388

1But surely the entire enterprise of quantum mechanics and science in general is predicated on there being causes to all events? Whether or not we can know the cause is an entirely different question. Statistical causes may still sufficient causes even if we can't know deterministically beforehand what will happen. – Jon Ericson – 2011-11-18T19:56:56.253

@JonEricson - That depends what you mean by cause. Something caused something from the set {X1, X2, X3, ...} to happen, but "nothing" may have determined which Xi was selected. – Rex Kerr – 2011-11-19T06:17:42.750

That seems an argument against determinism, not the principle in question. What I'm asking about is the view that events may have the null set of causes. ;-) – Jon Ericson – 2011-11-20T20:02:17.350

@JonEricson - What is an event? If I say that the-photon-passed-through-the-polarizer is the event, then it was necessary for the photon to hit the polarizer, but not sufficient. If I say that the-atom's-nucleus-ejected-an-alpha-particle, it was necessary to have an atom with a nucleus. Events with a null set of causes can't really affect anything that exists, or you can add in whatever the event affects as part of the event (otherwise you can't detect that it happened). That makes them rather limited in everyday experience. – Rex Kerr – 2011-11-20T22:19:04.237

So events with a null set of causes may occur, but they can't be detected so we can't verify they exist? That doesn't seem to be a very strong argument against the principle to me. I feel like I've fallen through a rabbit hole and can't get out... – Jon Ericson – 2011-11-21T01:53:45.187

@JonEricson - I just want you to be more precise with your definitions so that "event" is well-defined. It is very hard to reason accurately about something when you don't know what it is. I proposed a definition that I don't think works well, because it essentially tautologically forbids events with a null set of causes. Boring? Wrong? Propose something else! – Rex Kerr – 2011-11-22T11:57:49.450

I did my best to define event in the question. However, that could easily be a separate question (or five) on the site all by itself. What I'm curios about is if there philosophers who look at things that happen in the world and just throw up their hands: "We'll never know how that happened!" – Jon Ericson – 2011-11-22T17:28:15.733

1@JonEricson - With that clarification, I reiterate that your question is equivalent to asking about determinism, and determinism seems like a bad model given QM: if you ask why did that go left (as the event) the answer is just, because sometimes it goes left instead of right as an answer. That's pretty causeless, even if you can put numbers on the fraction left and fraction right, and it's predictable. – Rex Kerr – 2011-11-22T19:37:23.787

I didn't intend to answer the question, but this answer seemed to require a counter-argument and I don't think changing the question again is appropriate. It seems to me that we simply disagree on what the principle means and we don't actually disagree on much else. Maybe my full answer will help us decide what sort of disagreement we actually have. ;-) – Jon Ericson – 2011-11-22T21:01:03.290


It's difficult to know how an argument against the principle could begin. As humans, we seem to have a deeply ingrained model of the universe that implies all events have causes. Our instinct seems to be to assume a cause without having any explicit evidence that a cause can even exist. Consider the case of the the beginning of the universe. If any event is likely to be causeless, it is that event. And yet, there exist any number of theories that attempt to explain the Big Bang. Our intuition that everything has a cause seems to literally have no bounds—not even the universe can contain it.

One avenue of attack would be to suggest that our model of causation was itself uncaused and therefore it is not reliable. But that naturally leads us to question how we can make an argument against the idea of causation based on the principle of sufficient causes. (I am reminded of Plantinga's concept of defeaters here.) And more damaging, the argument, if it succeeds, merely shows that we can't trust our intuition, not that our intuition is false. It would be an attack on the epistemological question, not the metaphysical one.

We aren't asking about Determinism, which says that if we know the current state of the universe and the rules that govern it we can (in theory) know every other state of the universe. There are certainly good arguments against that hypothesis. And if we could find an argument against the "principle of sufficient reason", we could debunk determinism easily enough. (If things just happen, we can't very well predict them.)

But showing that determinism is a bad model for reality has no bearing on this question at all. If I find a coin on the ground with heads showing, there are any number of ways it could have gotten there. But since we all accept the principle of sufficient reason, we all agree that something must have caused the coin to be there and we all reject the idea that coins spontaneously appear on the ground. Nor is it a problem that the coin is showing heads rather than tails because there exist approximately equal number of causes that result in that state as opposed to the other. A coin carefully balanced on its edge excludes a number of causes, but we are certain that we will eventually find some set of causes that result in that state even if can never be sure which particular cause actualized it.

Quantum mechanics is a model of parts of the universe that suggests a number of counter-intuitive results, but as far as I can tell people who explore the model still expect to discover some set of causes for everything they observe. A simple test of that assertion is to imagine what will happen if a scientist notices something that the theory does not predict. They will likely redo the experiment, reinterpret their results, adjust the theory, or some combination of the above. What they won't do is say, "Oh well. Things sometimes happen that don't have any reason at all to happen."


There's no evidence that disproves the principle of sufficient reason (and precious little that proves it), so we can continue to behave as if it is true without fear of behaving irrationally.

Jon Ericson

Posted 2011-11-17T06:25:09.187

Reputation: 6 843

2You have not properly characterized quantum mechanics. Because of Bell's Inequality, physicists do not expect to find any reason for a particular choice of observable. They have simply expanded their definition of "reason" to include "picks at random from such-and-so probability distribution". – Rex Kerr – 2011-11-22T21:34:35.550

1@Rex Kerr: I fail to see how our discovery of quantum mechanics is materially different than our discovery of probability. A probabilistic theory of causation is still a theory of causation. But we seem doomed to be talking right past each other, so I think I'll just let this drop, if you don't mind. – Jon Ericson – 2011-11-22T21:49:12.227

You're missing the central point. Probability can be used either because there is a cause for each outcome but we do not know it, so we talk about distributions of outcomes instead; or because there is no cause for which outcome, only that there will be an outcome. But I agree that we are talking past each other, so I will stop. – Rex Kerr – 2011-11-22T21:56:06.347


If a cause is itself an event, every cause must also have a cause.

There cannot, then, be an uncaused cause but instead an infinite regress. If, on the other hand, there is an uncaused cause, the Principle of Sufficient Reason is false.

But perhaps a cause is not an event.

Jared Prince

Posted 2011-11-17T06:25:09.187

Reputation: 11

I made an edit which you may roll back or continue editing. You may see the versions by clicking on the "edited" link above. I think your point is critical: are (all) causes also events? I suspect they are not, which would allow one to have the principle of sufficient reason and also agents causing events without those causes being themselves events requiring causes. +1 – Frank Hubeny – 2018-09-02T01:29:23.887

Thank you Frank. If a cause is a living or other entity with volition, that cause results from either the will or an autonomic response. Autonomic responses of the kind that generally could be considered causes result from instinct or conditioning. The will results from the sum of our past experiences, interactions, personal preferences and genetic material or heredity. Whether the will or autonomic responses, they too seem to have causes though not easily discovered. If the cause is not such an entity, it too must have a cause. – Jared Prince – 2018-09-03T03:07:46.187

If there were an apparent first cause, it must have some kind of volition. If so, that must have come from something previous even if outside of our own space-time. If not, it must have been in stasis, so how did the stasis end without another cause? – Jared Prince – 2018-09-03T03:08:02.783

Sorry about the multiple posts, it's not letting me edit. If a first cause refers only to the genesis of our space-time and nothing external to it but that cause, I can see why such a stasis is not indicated. – Jared Prince – 2018-09-03T03:19:17.423

I think you made a good point with the suggestion that a cause may not be an event. That would involve some volition or will for it not to be an event. One might be able to say that the principle of sufficient reason does not apply to such agent-based choices but only to events. This is not to say that an agent-based choice does not have constraints or influences which might be events. Just that the choice itself is not an event nor completely determined. As a side note, I almost missed your comments. Use "@" with my name and this will send me a message. Again, welcome to this SE! – Frank Hubeny – 2018-09-03T04:29:52.047

@Frank Hubeny , I would think making a choice would be an event and acting on a choice would be an event even if either were observed only by the agent. I don't see how the distinction between "event" and "cause" relieves the agent from the Principle of Sufficient Reason. I would think the agent must also have sufficient reason to cause something i.e. intentionality. Intentionality too has its precursors, so it still seems an infinite regress though I acknowledge going back outside our own space-time may well be independent of PSR within our space-time. – Jared Prince – 2018-09-04T10:50:27.820

There is a chat room related to this topic that is still active: In brief I think an agent is the sufficient reason and if it is an agent the choice is not completely determined by prior non-agent events. Whatever the agent does becomes an event for other agents.

– Frank Hubeny – 2018-09-04T13:48:36.483

Induction is a mental process. If I understand correctly, Schopenhauer sees PSR as a mental principle rather than, necessarily, a principle of the world/universe. We are always trying to make sense of things even when they don't seem to make sense. Historically, we have been consistently wrong on some level in how we have made sense of things in the world. Nevertheless, being wrong didn't doom us (yet), as a species, and didn't deter our continued determination to find causes. Can we be sure that PSR is more than a mental principle? – Jared Prince – 2018-09-12T05:07:43.610