For empirical facts, it seems obvious that causality requires a time flow for the concept to make sense: A causes B implies that A happened before B.

Is it ever possible to have a causal relationship without a flow of time? Presumably this is impossible for empirical facts, as an effect preceding a cause goes against the laws of physics. Is this always true?

But what about relations of ideas? Is it possible to have a causal relationship between two ideas?

Colloquially we often speak of ideas having causal relationships:

  • Equation B is a consequence of theorem A.
  • Proposition Y is true because of proportion X.

Are these true causal relationships? Or is it merely us projecting our temporal thought processes onto the ideas and propositions we are analyzing and hence seeing causality where there is none? Can relations of ideas have causal relationships?

  • The term "cause" necessitates that one thing precedes another. But it sounds like you may be asking whether cause-and-effect are just illusions or coincidences.
    – user18800
    Commented Jan 4, 2016 at 22:21

8 Answers 8


Nancy Cartwright's definition, applied to mind-independent reality, does not reference time:

C causes E if and only if C increases the probability of E in every situation which is otherwise causally homogeneous with respect to E. (Causal Laws and Effective Strategies, 423)

The technical definition of "causally homogeneous with respect to E" is articulated on p423ff. I myself encountered the above definition in Cartwright's How the Laws of Physics Lie, which contains a slightly expanded version of the essay I've cited.

Switching to mind-dependent reality, you might want to consider using the term 'entailment', which can be seen as more general than 'causation'. Mathematical biologist Robert Rosen looks at the issue extensively in Life Itself, and uses the mathematics of category theory to rigorously describe the relationship between entailment in a model of reality and causation in reality. There is also some interesting discussion of whether causation/entailment exists in the model or also in reality in the beginning of Martin Hollis' Models of Man.

One way to consider the difference between logical entailment in formal systems and causation in reality is to recognize that not all conceptions of causation in reality are necessitarian. In contrast, logical entailment in a formal system is necessitarian. Given that there does not seem to be a link between necessitarianism and time, I shall say no more on this topic.


When you ask if "causality" requires an "arrow of time," I believe it is not impertinent to say, the opposite is more nearly the case.

The Newtonian, "billiard ball" picture of causation is reversible. Run the events backwards and you could not tell the difference. This is why such mechanical causality can be captured in mathematics, where there is no friction, disorganization, or entropy.

Entropy or the "Arrow of Time" is generally taken as the peculiar directionality of time as increasing "disorder" from some "point of view" and the apparent fact that such a point of view can never actually be eliminated from the picture. What occurs cannot be reversed. The "arrow" does not, cannot possibly, return to the bow. If this is "really" the case, and we assume it really is, then no "cause" is reversible or perfectly repeats itself, and "causation" appears pragmatic or nominalist.

You also ask if this is "just an idea." Teleology and the four Aristotelean "causes" seem to be making a comeback in "emergence," etc., against the old Newtonian, billard-ball model. If, say, "strange attractors" or "black holes" or even "desires" have some causal force, do they act before or after the "effect?" (Confusion of A series, B series time?)

Hegel seems to take a somewhat nominalist or at least ambiguous view of mechanical "causation." He rightly pointed out that causation is retrospectively selective. One cannot identify a "cause" until one has identified the "effect." So in conscious representation, effects always precede causes.

So, I guess my answer to your question is: no.

  • I didn't think of the first part, interesting. However I didn't say that "it's just an idea", but whether causal relationships are possible between ideas Commented Jan 5, 2016 at 16:39
  • Yes, apologies. I was not paying close attention to the second part, which deserves an entirely different answer. But again, this presses the definition of "causality." Can one "idea" necessarily "determine" another? This is the stance of rationalism. Hobbes, for example, was struck by the idea that a Euclidean proof could "compel" him to assent. Today the question would probably be reduced to psychology. Can one "idea" or "image" force the association of another, as in so many behavioral experiments? Commented Jan 5, 2016 at 18:04
  • ....Perhaps, but the very notion of reason and freedom, in the Kantian sense, rests upon the independence of "ideas" from mechanical causation. Our transcendental will presumably escapes, delays, reverses, or redirects "natural" causes and effects. Kant called this a second type of "causality" and set out to demonstrate its internal necessities. However, the role played by "time" in such conceptual orders is highly debatable, even obscure. Certainly "before" and "after" are not so clear in consciousness, as Hegel noted. Commented Jan 5, 2016 at 18:13

The answer depends on the definition of "causal", but traditional metaphysics did not require causal relation to be temporal. The most famous counterexample is the relation between God and the world in Christian theology modeled on the relation between One and the world in the neo-Platonism of Plotinus. God or One are eternal/atemporal and do not even exist, their being logically precedes existence and time that flow from it, i.e. they cause the temporal world to exist. Kant modeled the relation between his atemporal self, which exercises free will, and empirical self inserted into phenomenal causal chains that play out in time on this theological causality, but he would not use the word "cause" in this context.

There is an interesting application of this atemporal causality in modern physics, the Hartle-Hawking no-boundary cosmology misrepresented in popular literature as having atemporal piece "before" the Big Bang. Butterfield and Isham in On the Emergence of Time in Quantum Gravity explain that this is not the case:

"One often sees a picture in which a cone-like spacetime structure (representing a cosmological solution of classical general elativity) is attached to a spherical shape that represents a Euclidean 4-manifold. This erroneously suggests that the bottom sphere is straightforwardly earlier than the classical cosmology represented by the open cone in the top half of the figure. But the 4-manifold is not earlier: there is no temporal relation between the two halves represented in the figure (or their parts)!"

The relation is more akin to the theological causality with God replaced by a quantum ensemble of Euclidean 4-manifolds, and the world by Lorentzian spacetime of Big Bang cosmology implemented by their superposition. This is often misleadingly expressed as "tunneling from nothing", with "nothing" referring to the Euclidean "pure space", and "tunneling" to "moving from a time variable that is a real number to one that is purely imaginary". An incomplete classical analogy would be atoms in a solid "causing" elastic waves when they follow a periodic pattern of motions. The relation between motions and waves is not temporal, they are co-extensive, but in the no-boundary cosmology the time itself emerges from an atemporal substrate, just like waves from atoms.

While temporality can perhaps be avoided causality does seem to require some sort of partial order relation to be meaningful. In the cases of creation and emergence time is replaced by a "flow" between different "levels" of ontology. This being said, even if one admits atemporal causality it does not have to apply to every case of logical entailment. Indeed, on the Platonist reading there are no causal relations between ideal forms, they are at the same level, independent and co-eternal, hence mathematical derivations do not represent any kind of causality. Causality of derivations would not conform with the usual mathematical intuition either, if we have two equivalent mathematical facts (e.g. being even and being a sum of two primes) what sense would it make to say that one is caused by the other?

  • "replaced by a "flow" between different "levels" of ontology" fair enough. But then how does one differentiate between "artificial flow" imposed by our thought process, vs "inherent flow" which indicate a true causality? Commented Jan 5, 2016 at 0:24
  • 1
    @Alexander S King The "flow" between different levels may well be "inherent" as well, certainly theologians view creation from nothing as independent from our thought processes, I think the same goes for Hartle-Hawking. According to Kant, even temporal "true causality" is imposed by our mental faculties, and he agrees with Hume that there is none other to be had. I think the distinction between inherent and imposed depends on metaphysics one adopts, and so will the methods of differentiation. To empiricists the question is moot except within a model describing "reality" and "observer".
    – Conifold
    Commented Jan 5, 2016 at 1:16

The issue is controversial. In particular, imagine we live in a universe with cyclic time, or imagine that the geometry of space time is such that some processes go backward in time. Then causes could precede their effect. There are also interpretations of quantum mechanics that involve retrocausal effects.

However we have deep intuitions that causes precede their effects. A simple argument to express that intuition is that if effects preceded their causes then the cause could be undone and their wouldn't be a causal relationship so we'd have a potential contradiction.

Regarding mathematical statements it is indeed a projection of our thought process on deductibility relations: there is no causal relations between premises and their consequences, but the fact that we use causal language is an interested feature that were analyzed by some authors (sorry I don't remember the names).


I think the "problem" or question here is essentially confused by words (or even techniques for solving problems) that push the answer one direction or another, if they come close to answering the question at all. The first part of the problem is a circularity or loop that cannot be avoided. The circularity relates in a way to the problem of defining words, and to the nature of reference to a limited circumstance: as humans we experience the world and universe in a certain way, and there is no way to escape that. This a problem that plagues dictionaries and mathematical systems. That is one problem.

The second issue is that when we as humans look at space-time we can easily visualize an empty space (or at least we think we can), and we have words (terms) that seem to work for that (Empty? Nothing?), but we (apparently) cannot, or easily, imagine a "timeless" space or the absence of time. We may think we have words for that, but we probably do not. Eternal and infinity may look promising as useful terms, but are they? Saying something or some condition is timeless means that something exists through time, not that it exists outside time, or there is no time. Avoiding that is not a simple issue. Using infinite or derivations of that word about time suggest something else, and there is no way to say that is helpful.

It is pointless to consider causation without the effect of time. Causation in any useful physical sense does not just suggest time, it requires time, and not just because of words like "it follows that..."

For that matter, the word causation, and its usual accepted meaning, may be part of the problem.

The early attempts to deal with this time issue in mathematics (Liebnitz and set theory) strike me as saying more about the theory used than about the problem itself. The theory does not come close to resolving the issue. More recent approaches to the problem suffer the same fate. The solutions suggest an image of trying to demolish a large modern concrete and steel structure the size of a mountain using only a pick and shovel, or scissors. The tools work fine for doing what they were meant to do, but the problem is beyond such tools. Even such a suggestion minimizes the nature of the problem.

What this may mean is only that there may be no "first cause" or any cause to the existence of space-time (our universe) as we can conceive of it now because without the matter that makes up space there was no time. And this obviously relates to whether our human idea of "time" has to be better defined and resolved. It does not mean we cannot continue researching and learning about the origins of the universe, and at some point the nature of the issue and problem itself may be more clearly revealed.


One has to distinguish between ideas and events. Both live in different ontological domains.

  1. Causal relations may hold between events. A series of events makes up a fact.

  2. Logical relations may hold between ideas or propositions. The concept of time does not apply to ideas.

A causal relation like "Event A" causes "Event B" presumes that A is earlier than B, more general, that B is in the forward lightcone of A - at least in the usual physical word models.

  • What you say makes perfect sense, but I am unclear on the role of time here in, say, a Kantian-type scheme. Since space and time are both a priori and categorical prerequisites of concepts, would that imply that concepts and "logical order" do have some sort of underlying, necessary" time" order? This is clearly so in Hegel. But I'm not sure how Kant relates the givenness of "time" to the logical process. Perhaps it isn't entirely clear in Kant himself? Commented Jan 5, 2016 at 18:24

There is a strong counter-example in quantum physics which is causing a lot of headaches. Entanglement is a relationship between two particles, in which some property of each is described by a single wave equation. For example the polarizations of two photons may be entangled. Such quantum properties famously exist in an indeterminate state until they are measured, at which point they collapse into the measured state. This means that if you measure the state of one entangled photon, this collapses the wave and immediately determines that of the other photon. By immediately I mean with no measurable delay, certainly propagating far faster than the speed of light.

Consider an experiment in which two entangled photons are sent towards different detectors. When the first one arrives its measurement, event A, instantly collapses the state of the other and pre-determines what its detector will register as event B.

But another observer in a relativistic reference frame might well see B occurring before A. This is not some trick of time dilation to be ironed out by a little arithmetic, it is a genuine difference in the time coordinate grids of the two observers. Neither has primacy. This creates a paradox; To one observer, A caused B. To the other, B caused A.

Worse, the original experimenter may choose which state to force A to measure. B then confirms the entanglement. So one observer now sees B causing the event which the experimenter thought they had caused, by implication even causing the experimenter to make the choice they did. Various delayed-choice experiments have indeed been carried out and the paradoxes confirmed in some detail over many years of experimental refinement.

This appears to be telling us that the relationship between time and causality is not absolute but relative, just like simultaneity. The cost of maintaining temporal causality is to discard some other equally fundamental assumption about reality, of which there are several to choose from.

Headaches all round.


I think you could merge both themes with the new concept of Entropy (information entropy).

In physic, causality means that some event at time t depends only in events before or at time t. This has to do with invariance and conservations, since conservations exist, it guarantees that eny new state is a simmetric transformation of a previues state that has invariance on the conservative magnitude.

In ideal concepts and logic, entropy will yield in information. in order to get a conclusion you need to assume some information before, and any new conclusion over the priors actually will be the same meaning with a different point of view. It just is not evident for us but in information view it actually is the same concept.

e.g: "Equation B is a consequence of theorem A." then 'equation B' is actually a transformation equivalent to 'theorem A' that conserve the same information but we project the same information in different ideas as our language let us map in.

I didn't use the entropy concept as i wish but i hope it gives some new ideas.

I would conclude that causality in physics or logical thinking is actually a conservation of some variable. The most common conservation is over time, so we usually associate invariance conservations with time and hence with causality.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .