APA 2014-2: Against causal reductionism

relativityby Massimo Pigliucci

Second report from this year’s meeting of the Eastern Division of the American Philosophical Association. This session, under the general heading of philosophy of science, was actually constituted of just one talk, entitled “Against causal reductionism” and delivered by Chris Weaver (Rutgers University) (the session was chaired by Michael Hicks, Rutgers University).

As you’ll see from my interspersed and final comments, I actually disagree with the author’s main thesis, and even the way he goes about defending it. But the talk did stimulate my neuronal firing, and I hope it will generate some thoughtful discussion here, as well as provide another example of what professional philosophers do at their daily job.

Weaver began by observing that the consensus view in physics is that the universe began in a very unusual state of very low entropy, but maintains that the characterization of such initial state as “unusual” is based on the highly questionable principle of indifference [1] and is otherwise difficult to defend. I don’t really have a bone to pick in that particular fight, and it honestly wasn’t clear to me how exactly this related to the main topic of the talk, causal reductionism, but that’s where we are going next.

Weaver thinks that causal reductionism is based on an unsound argument from physics. Causal reductionism – in this context – is the view that causal interactions reduce to non-causal facts, that they are nothing above and beyond lawfully related events, where natural laws themselves are not “causal.” (Of course, there is a huge literature on the very concept of causation itself, but we’ll leave that for another day. [2])

Causal reductionism, according to Weaver, is not implied by empirical analyses of causation but depends instead on the following “argument from physics”:

P1) Physical science only requires natural law and physical history (no causation)
P2) If 1, then causal reductionism holds
C) Causal reductionism holds

Therefore, according to Jonathan Schaffer, among other authors, “causation disappears from sophisticated physics.” Weaver doesn’t buy it, and proposes to attack the first premise above, by arguing that there is a distinction between the formalism and the interpretation of any given theory T, and that while some particular physical theory T lacks the formalism of causation, any sensible interpretation of T will have to include causal talk.

Take the idea of a gravitational field in the general theory of relativity (henceforth, GTR). While the formalism of the theory (i.e., the equations) doesn’t include any talk of causes, the field itself has to be interpreted – according to Weaver – as having causal properties: if gravitation is the conformation of space-time itself, then obviously it has causal properties.

GTR is based on four principles: relativity, general covariance, finitude of the speed of light, and equivalence [3]. Weaver maintains that the principle of equivalence (between gravitational and inertial mass) includes language concerned with the effects of gravitation, i.e., it is framed by using talk of causality. Even when it is not expressed using causal language per se, the principle is usually accompanied by language that implies causal notions (for instance in Sean Carroll’s 2003 textbook [4]). Indeed, Einstein himself understood the principle of equivalence in causal terms, which Weaver demonstrated by using a number of direct quotations.

So, the idea is that although the field equations of GTR do not explicitly include any causal talk, one cannot avoid but interpret them as telling us about the behavior of the gravitational field, including, therefore, its (causal) effects on the motion of objects and free particles.

Weaver appropriately quoted a number of experts on GTR confirming his interpretation that Einstein did talk of the theory in interpretive causal terms, whenever he was discussing how the theory accounts for the inertial motion of bodies.

(At this point in the talk I had the perhaps obvious thought that while all of the above is fine, GTR is still a classical theory, and that causal talk – and hence Weaver’s approach – still break down when we get to quantum mechanics, or to a future quantum theory of gravity. Turns out the author himself dealt with this objection toward the end of his talk, not at all in a satisfactory manner, I think.)

Yet another example of causation within GTR is the theory’s prediction of gravitational waves, which do causally affect both fields and matter. The recent (alleged) discovery of gravitational waves [5] was the result of observation of effects of such waves on the cosmic background radiation, and moreover, gravitational waves are certainly emitted by causal interactions.

Weaver maintains therefore that causation enters sound physical science by way of a proper understanding of general cosmological theories such as GTR. After all, spacetime points, according to GTR, are associated with “domains of influence” that are almost universally interpreted causally.

Crucially, GTR does not reduce the above causal influence talk to anything more fundamental or primitive, regardless of mistaken (according to Weaver) identification by some authors of causal influence structure with light cone structure [6].

Finally, as it is customary structure in many philosophy talks, the author considered (and attempted to rebut) a number of possible objections to his thesis.

The first such objection notes that the dynamical laws of GTR are time reversal invariant, which means that any causal reading of those equations implies a denial that causes necessarily precede their effects.

Here Weaver surprised me by simply shrugging the objection away! The principle that causes must precede their effects, he maintained is false. What he was referring to was the possibility of simultaneity of cause and effect, which has been proposed by others. The problem is that this would require a radical principle of instantaneous action at a distance, of the kind that Newton was worried about, as Weaver himself was keenly aware. The best he had to offer was the claim that instantaneous action at a distance may not actually be the case, but that the mere possibility is not incoherent. Okay then, but I thought we were talking physics, not logic.

Worse, the point about the time reversal invariance of GTR equations would also imply backwards causation, which Weaver accepts (well, he has to!) on the grounds that GTR does not preclude closed timelike curves [7], which in turn would make time travel possible [8].

Weaver also nodded toward the problem with quantum mechanics that I raised above, but said, and I quote: “well, you know, quantum gravity is a mess…”

A second objection points out that GTR is not itself a fundamental physical theory, so the causal reductionist should not be worried, at the least not yet. GTR will eventually have to yield to quantum mechanics in ways that would rub out any attempt to understand the causal activity of the gravitational field as fundamental physical activity.

To this Weaver, rather astoundingly, simply bit the bullet and acknowledged that okay, so maybe fundamental physics can do away with causality, but not all physical theories can. But this is worse than a Pyrrhic victory, I think. To begin with, because it is well established in philosophy of science that “special sciences” (i.e., everything but fundamental physics, including non-fundamental physics) do effectively deploy causal talk and cannot, apparently, do without it [9]. The puzzle of causal vs non-causal talk in science has always been at the fundamental level. If it turns out that the only reason GTR has to engage in causal talk is because it isn’t a fundamental theory (and we know it isn’t!), and that once we move to more fundamental levels of description and explanation causal talk yields to nomological talk (i.e., talk in terms of laws of nature) then the game is up for the critic of causal reductionism. Boy would I love to hear about this from physicists and philosophers of physics!

Massimo Pigliucci is a biologist and philosopher at the City University of New York. His main interests are in the philosophy of science and pseudoscience. He is the editor-in-chief of Scientia Salon, and his latest book (co-edited with Maarten Boudry) is Philosophy of Pseudoscience: Reconsidering the Demarcation Problem (Chicago Press).

[1] Principle of indifference, Wiki entry.

[2] See: The Metaphysics of Causation, by Jonathan Schaffer, Stanford Encyclopedia of Philosophy.

[3] Principle of relativity; General covariance; Equivalence principle.

[4] Spacetime and Geometry: An Introduction to General Relativity, by S. Carroll, Addison-Wesley, 2003.

[5] Gravitational wave discovery looks doubtful in new analysis, by Clara Moskowitz, Scientif American, 22 September 2014.

[6] Light cone, Wiki entry.

[7] Closed timelike curve, Wiki entry.

[8] Unwinding time, by S. Carroll, The Wall Street Journal, 17 December 2011.

[9] See, for instance: Every Thing Must Go: Metaphysics Naturalized, by J. Ladyman and D. Ross, Oxford University Press, 2007.


Categories: commentary

Tags: , , , ,

55 replies

  1. Marko: No, physics never uses the term “spontaneous” to be equivalent to “not being induced by a cause”. I said that these anti-causality arguments were a century out of date, and sure enough, you give me a reference to 1916! Exactly 99 years out of date. If you are going to try to understand QM causality, you have to look at something at least 10 years later.

    You say that “QM the probability is a fundamental property of nature, independent of our ignorance.” That is true in some interpretations of QM, and not others. All of the interpretations give causal explanations. (Some more successfully than others.) Probability is certainly part of many of those causal explanations.

    While interpretations of probability vary, I do not think that you will find any QM textbook that endorses your view that quantum probability contradicts causality, or that spectral lines are only partially causally explained to the extent that they are influenced by a deterministic theory.

    For an explanation of how probability in QM can be entirely about our ignorance, see Quantum Bayesianism or this 2014 Nature article.

    Dominik: The same causal reasoning goes into EM theory today. The popular EM textbook by Griffiths says that the principle of causality is “the most sacred tenet in all of physics”. While Maxwell’s equations are formally time reversible, the reversed solution is nearly always unphysical for some reason.

    Likewise causal reasoning goes into GR. Marko says it is put there by hand, whatever that means. It is part of GR as presented in textbooks.

    Your argument about waves being emergent and hence causal does not make any sense in GR. If it did, you could apply it to my simple example. Does the Sun’s gravity effect the orbit of the Earth? I say that the answer is obviously yes, and it does not matter whether you think of waves as emergent, or anything like that.

    According to a common view, waves are the most fundamental things there are. All of gravity is just a lot of waves. So basing an argument on waves being emergent is not going to get you anywhere.

    The argument about causality being related to conserved quantities does not make any sense either. It is true that causality and conserved quantities are used all the time in physics, so a naive non-physicist might think that there is some relation. Explanations in physics often involve both, but there isn’t really any direct relation.


  2. schlafly,

    I let this comment pass, but barely. While you do make interesting points, it is obvious to me that so do your interlocutors (including, humbly, myself). Therefore phrases like “does not make any sense,” with which you liberally and unnecessarily pepper your comments are entirely uncalled for. Why don’t you try to simply address the specific points others raise while holding back your obvious contempt for anyone who disagrees with you, especially (but apparently not only) if they are philosophers? Thanks.


  3. Hi Marko,

    … the theory [GR] itself does not distinguish between cause and effect, we put this distinction into the theory from the outside, i.e. from observation.

    But the observations are of how nature actually is, and if we look at nature we see a direction in time and we see causality. So observation/nature is telling us to put a cause/effect distinction into the model.

    This is not only about thermodynamics and the second law. For example, we can detect the cosmological redshift without invoking thermodynamics, and there would still be such a shift in an adiabatic (dS = 0) expansion/contraction.

    It thus seems that nature does have a direction to time at a very fundamental level (i.e. it is not just emergent at a higher level, owing to thermodynamic irreversibility). Perhaps that is not encapsulated in theories of GR and QM, but then they are descriptions of *some* aspects of fundamental reality, rather than being complete accounts.

    I’ll also repeat my earlier point that a mathematical theory/description of a physical entity is not the same thing as that physical entity, Tegmark notwithstanding, the chief difference being that entity itself has the capability of bumping into something else and thus of `causation’. Afterall, what is the point of an electron if it cannot cause? (Or do people really want to go the full Tegmark?) Personally I’d say that the capacity to “cause” is the chief attribute necessary for something to “exist”.

    Most likely this fundamental directionality to time is wrapped up with causation, and for that reason the idea that fundamental physics dispenses with causation doesn’t ring true to me — but, as has been noted, there has been no clear account of what the concept of causation actually is, so I’m still unclear what the claim of the absence of causation actually amounts to.

    5th and thus last comment, but I’ll read any replies with interest.


  4. Schlafly,

    This will be my last comment to you, and I’ll address only this:

    Likewise causal reasoning goes into GR. Marko says it is put there by hand, whatever that means. It is part of GR as presented in textbooks.

    The ultimate authority on what is (or is not) a part of a theory are equations, not textbooks.


    This is not only about thermodynamics and the second law. For example, we can detect the cosmological redshift without invoking thermodynamics

    Given a process of measuring the redshifted spectrum of a distant galaxy (using say a telescope and a spectrometer), let me illustrate what the time-reversed process looks like.

    The process begins by spectrometer emitting a photon through a telescope, which then begins its journey towards a distant galaxy. As the space between the telescope and the galaxy contracts, the photon is blueshifted when it reaches the galaxy. Once there, it is being absorbed by a star, where it hits a helium nucleus and causes fission into two hydrogen nuclei. The fission spends the energy of the photon and some of the surrounding particles, cooling the star down. Since there is net influx of such photons into the star from all directions, the star eventually cools down into a diluted hydrogen gas, which then expands away due to the kinetic energy of every molecule being large enough to break orbit from the rest of the gas.

    First, note the causal chain: spectrometer causes the photon to be emitted, gravity causes it to be blueshifted, and the photon then causes fission of helium into hydrogen in a star, which consumes energy and causes the star to cool down.

    Second, note that in the above scenario no laws of physics have been violated, except the second law of thermodynamics. This can be tracked down by precise accounting of the entropy of the star and of the spectrometer at the initial moment (emitting of photon) versus the final moment (fission of helium).

    It thus seems that nature does have a direction to time at a very fundamental level (i.e. it is not just emergent at a higher level, owing to thermodynamic irreversibility). Perhaps that is not encapsulated in theories of GR and QM, but then they are descriptions of *some* aspects of fundamental reality, rather than being complete accounts.

    Time direction is indeed very fundamental, but it is *precisely* due to thermodynamic irreversibility, as illustrated by the above example. Also, indeed it is not encapsulated in GR and QM, and that makes them incomplete, short of postulating second law of thermodynamics as an additional axiom of the theory. However, people feel unsettled when a fundamental axiom applies only to a large (and unspecified!) number of particles, and thus the arrow of time is considered a *problem*.

    I’d say that the capacity to “cause” is the chief attribute necessary for something to “exist”.

    I’d say that the capacity to “interact” is the chief attribute for existing. At the fundamental level, interaction has no time-directionality, while causality does. Second law of thermodynamics notwithstanding, all interactions work equally well for both orientations of time axis, and the concepts of “cause” and “effect” should be in the eye of the beholder. 🙂 But the second law of thermodynamics breaks this symmetry, and our intuition about causality was built on it.


    I think this is my fifth post, so I’ll just say it was a pleasure to discuss this stuff! 🙂


  5. Aravas,
    My point is that supposedly unfathomable quantum concepts, such as the uncertainty principle, are applicable to the human scale and understandable, with a little thought. The problem is the tendency to try to reduce everything to static terms and measures. This is because form is what we can describe. Whatever the form that energy takes, wave, velocity, mass, etc, is defined and understood as form. Yet it is only the energy which is conserved, while the form is transient.
    This goes to the issue of time I keep raising; That the sequence of events and thus the narrative progression from one to the next, is an effect of the dynamic process creating and dissolving this form. So it is energy which truly manifests what we call the present and the process turns potential into actual and then residual. The form goes future to past.
    Think in terms of a movie projector; We think of the movie as a sequence of events, while what we are perceiving is the projector light that is a constant and the frames pass by, from future to past. Just as our consciousness is always in the present, while the thoughts it manifests coalesce out of potential and dissolve into residual, ie. future to past.
    The arrow of time for energy goes past to future, while the arrow of time for form goes future to past.


%d bloggers like this: