Monday, March 13, 2017

A dilemma for elementalism

Teaching history of philosophy always produces in me a lively sense of contingency in intellectual history. One example stands out: Leucippus and Democritus proposed an atomic theory around 440 B.C. This striking proposal ran into the sand, however, and never produced a viable research program.

Why not? All you need, technically speaking, are devices for measuring and weighing small amounts of material fairly precisely. The ability to do that was not significantly greater in the early 1800’s when Dalton and Avogadro succeeded in reviving atomism.

I used to think that the Greek atomists just had the rotten luck to live at the same time as two of history’s greatest philosophers, Plato and Aristotle. Atomism, though right, lost the dialectical battle.

I now think that western philosophy tacks between two strategies for explaining What Goes On.

One strategy explains the properties and behavior of entities by appeal to composition: the properties and behavior of the fundamental objects that make the entity up. Call this strategy ‘Elementalism’.

The other strategy explains the properties and behavior of entities by appeal to their kind: and a full understanding of the essential properties of the kind will explain what the entity does. Call this strategy ‘Formalism’.

Formalism dominated from 400 BC until 1400 AD. Our course in early modern philosophy is the story of its overthrow by elementalism in the Scientific Revolution.

Either strategy yields philosophical problems, problems which the other strategy tends to solve – at least for a time.

The main problem besetting any elementalism is composition.

Matthew hits a baseball toward the house and breaks a window. [1]

Elementalism explains this by the bonds which hold atoms together into molecules: the molecules forming the baseball (B) and the window (G). The velocity of B on G is sufficient to disturb the molecular bonds of G.

This explanation makes no reference to the baseball, the window, or to Matthew. What matters for causal explanation is what goes on at the level of B and G. Baseballs, windows, and teenagers with poor judgment may as well not be there. Since we need not countenance them in order to make our best (elementalist) theories true, we can be anti-realists about macro-objects.

If we wish to believe in baseballs and grandchildren – and we do – we could suppose they supervene on the fundamental particles. But then they're epiphenomenal: their presence makes no difference in the world.

If they’re to make a difference, we seem to be committed to saying not only that “Events involving B cause events involving G,” but also “Events involving the baseball cause events involving the window.”

But that gives us two causal relations: double determination. What goes on is over-oomphed.

Indeed, since we have many levels operating:
  • atomic
  • molecular
  • macro- (baseballs and windows)
  • physiological (Matthew’s hitting the baseball)
  • neural (events in Matthew’s brain causing his motor movements)
  • mental (Matthew’s ill-considered decision to hit the ball toward the house)
what goes on is umptly over-oomphed.

Here’s the dilemma.


The higher-level events purportedly involving higher-level entities are not, strictly and literally, real events at all, or derivatively real.

But then they would be entirely epiphenomenal – not really part of the causal story of the world.


They are real, with their own causal relations.

But then ‘what goes on in the world‘ is over-determined.

In short, the constant dilemma with elementalism is that it threatens to give us either too little reality (no grandchildren) or way too much. Indeed, since there may be no limit to the number of levels at which causal explanation can take place unintelligibly too much.

But we thought elementalism was superior as an explanatory strategy.

Does formalism do any better? Arguably, yes.

One promising formalist strategy currently emerging is a revival of Aristotelian-style Hylomorphism. Robert Koons gives a nifty summary of the current state of this research project. [2]

On Koons’ version, the causal powers of a substance’s parts become powers of the whole substance, and so the over-determination problem doesn’t arise:

“(I)f I stand on a scale, is it I (as a whole) or my parts (collectively) that cause the pointer to move? If the powers associated with weight have migrated from my proper parts to me, my weight can be the unique and non-redundant cause of the scale’s response.” (Koons 2014, 8).

Tom Pyne
Department of Philosophy
Sacramento State


  1. Tom, thanks very much for this. Do you think you could explain to me what exactly is wrong with overdetermination as you are using the term? Will it lead us to make inaccurate calculations or false predictions?

    I think a standard kind of response today to what you are are arguing is that these are just different explanatory frameworks, providing different ways of describing the same physical systems. It is possible to use the term 'real' in such a way that the only entities that are real are those acknowledged in our most fundamental theories of the world. But I don't see anything good coming from that. To me the better use is to acknowledge as real anything that has real explanatory value and does not conflict with predictions made by those theories.

    Acknowledging the existence of baseballs, windows and grandsons in no way conflicts with quantum field theory or particle physics. and doing so increases our explanatory power. There are other entities, e.g., disembodied minds, that do make different predictions than our most fundamental theories, and that is a reason for thinking they are not real.

    1. Randy,
      Thanks for the helpful comment. Another of those situations where I don’t know whether we disagree or not!

      What’s making me unsure is your juxtaposition of two phrases:
      • ‘different explanatory frameworks’
      • ‘providing different ways of describing the same physical system’.

      If the explanatory frameworks are simply descriptive, then you’re right: the dilemma does not really arise. Using a different level of description (“the baseball broke the window”) would be as harmless as saying that the sun ‘rose’.

      The truthmaker of ‘the sun rose’ is really the earth’s rotation, and we can construe the description as not inconsistent with that.

      How are you using the phrase ‘the same physical system’? If the physical system is B and G, then what’s really going on is going on at that level. In that case, calling the baseball and window ‘real’ is kind of honorific.

      On the other hand, if the physical system is B and G and the baseball *and* the window, and if both pairs have explanatory value, it’s hard for me to see how that’s not over-oomphing.

      One could maintain the reality of baseball and window by showing a supervenience relation to B and G. But this makes baseballs, windows and grandchildren epiphenomena in the causal story of what goes on in the world. I want more than that.

      So it seems that the dilemma for elementalism remains.

      “(A)cknowledging the existence of baseballs, windows and grandsons in no way conflicts with quantum field theory or particle physics…” makes a point – and a good one – about logical consistency. I’m talking metaphysics.

      To the extent that current hylomorphism is a program for ascertaining the relations between wholes and their elements that avoid this dilemma, it seems to promise a genuine clarification of a scientific view of the world.

      The question will not be settled by the science alone. (But you knew I’d say that!)

      As to your last point, yeah. We’d have as hard a job explaining how disembodied minds can be in intentional states as we would explaining how we human organisms can be. Indeed, it would be the same job.

  2. Tom, thanks, that is helpful. (Forgive the prolix reply.)

    I'm still not sure I understand what makes overdetermination bad. Clearly, there is no logical inconsistency in saying, as you say above, that: “Events involving B cause events involving G,” but also “Events involving the baseball cause events involving the window.”

    You indicate that the problem lies in the fact that there are two causal relations, but it’s not obvious to me that this is a problem.

    I hope this is a generous restatement of your argument: Let’s assume it is true that there is a causal relationship between B and G and that the causal powers attributed to B are sufficient to produce G. Now we add the assumption that the baseball causes the window to break. But since B is already sufficient to produce G, then the causal powers of the baseball cannot be making a difference. Since the ability to make a difference is what underwrites our claim that the baseball is real, then we are forced to deny the reality of baseballs and windows, which we all agree is absurd.

    We see a version of this problem arising with mind/body dualism, where the problem with irreducibly mental causal events is that once we acknowledge the causal sufficiency of neurological processes, irreducibly mental causation makes no difference. The same kind of problem would arise for anyone who continued today to insist on the existence of an elan vitale or of impetus as a causal force over and above inertia. And, of course, exactly this problem arose for our understanding of the causal powers of God when the notion of a fully mechanistic self-sustaining universe gave birth to the Newtonian period.

  3. (continuing)

    The reason I think these problems are different than the one you are raising is that they are unequivocally committed to the existence of distinct causal processes. But in your example, we have the option of saying that they are exactly the same causal process, differently described.

    I accept your suggestion that for something to be real it must be capable of making a difference. My way of saying that would be that it participates essentially in a type of explanation that makes a distinct contribution to our understanding of the world. But to say that it makes a difference does not entail that the nature of this difference can’t be described differently. In brief, the power of B to cause G is just as susceptible to redescription as B and G itself. The appearance of overoomphing is, in this case, an illusion.

    In your reply to me, you asked: “How are you using the phrase ‘the same physical system’? If the physical system is B and G, then what’s really going on is going on at that level. In that case, calling the baseball and window ‘real’ is kind of honorific.” My answer to that question is that the physical system is what it is. “B causes G” and “Baseball breaks window” are two different ways of describing that process. There is (in quantum field theory) a description of that process that we would call our most fundamental description, but this just means that this is our most comprehensive description of reality, not that quantum fields are the only things that are real.

    The stubborn reality we are dealing with here, something I’m sure we both accept, entails that our most comprehensive description of it just can’t provide all the explanations we require. But we understand exactly why that is the case, viz., that we, and even the most powerful computer we can imagine, are swamped by the computational complexity of the task at that level. This is why we need to ascend to higher levels of description to get the explanatory tasks done. We see this occurring even at microlevels. Our concept of a subatomic particle, e.g., is a distinct way of describing the reality that is described even more fundamentally as fluctuations in quantum fields. If you didn’t care about being comprehensible, you could have written your post as a problem about acknowledging these two levels of description.

    (But, with this interesting problem: In this context causation is not even a thing. Ultimately what we are calling a causal process is a description that is replaced at more fundamental by purely statistical relationships. So if you were to try to raise the problem at this level you would have to do it without reference to causation at all.)

    Perhaps I am just a few years behind you, but right now I continue to subscribe to your earlier view. The loss of Democritus’ writings may have been the biggest intellectual tragedy of the ancient period. Elementalism is the most brilliant and penetrating explanatory strategy ever conceived. Formalism is philosophically important insofar as it may help us to develop the optimal way of describing the way that different levels of description and explanation relate to each other, but it probably does not hold the keys to distinct future scientific achievements.

  4. Your restatement of my argument is not just generous but accurate. But it is not (alas!) my argument. I borrowed it from Trenton Merricks, who was cited in my post. However, for Merricks it’s not a dilemma for elementalism but an argument for it.

    Since macro- objects like baseballs and windows would yield double causation, ‘over-oomphing’, there mustn’t be any.

    So not everyone would agree that denying the reality of baseballs and windows is absurd!

    As Merricks himself puts it, “There are no statues or chairs or rocks or planets. But there are microscopic objects. Let’s call them—whatever they may turn out to be—‘atoms’. And although there are no statues, there are atoms “arranged statuewise.” There are also atoms arranged chairwise, atoms arranged rockwise, and atoms arranged planetwise.”

    As I once heard Wesley Salmon remark, ”One man’s counterargument is another man’s modus ponies.”

    So Merricks solves the composition problem, which I claim besets elementalism, by denying the reality of composition.

    Your solution, to deem the different levels, B and G vs baseball and window, merely different levels of description devoid of any ontological freight, constitutes another way to avoid the composition problem. The difficulty I see with it is that it opens two possibilities, neither of which I think you would be happy with.

    1. The metaphysical commitments of all the levels of description and explanation really only get cashed in on the fundamental level. You make the point, “There is (in quantum field theory) a description of that process that we would call our most fundamental description, but this just means that this is our most comprehensive description of reality, not that quantum fields are the only things that are real.” But really, that’s exactly what you’d have to say.

    2. It treats the issue as one of heuristics: different levels give us different help in understanding, but no general theoretical strategy for understanding the reality of composition itself. Yet this seems necessary, since Merricks and Koons (and Mayes and Pyne) agree on the science, yet disagree on the reality of composite objects.

    And it’s there that the argument becomes an argument for an alternative strategy: what I called ‘formalism’, one version of which is ‘hylomorphism’.
    And it’s here where I’m going to send you up the flue, I suspect.

    Maybe – maybe – we should entertain the possibility that the elementalist research project has now itself begun to flag. It produced the greatest intellectual gains in human history, explaining what happens by appeal to smaller and more elemental bits. Perhaps we are approaching the limits of this strategy. After all it can’t be atoms ‘all the way down’.

    So perhaps the path is clear for a revival of hylomorphism to enlarge our toolkit for understanding how macro- composite things are ultimately composed of micro- things while acknowledging the reality of both, producing a consistent multi-leveled ontology consistent with our scientific knowledge.

    Couldn’t that be a key to future scientific achievements?

    A third strategy has become available to us, of course: seeing how things fit together in probabilistic Boolean networks. This now possible because of increases in computing power. What little I understand about this new strategy comes from Edward Dougherty’s very readable book ‘The Evolution of Scientific Knowledge’. But it seemed to me that this new-found power to study dauntingly complex systems is just what a disciplined hylomorphism needs.

  5. Modus ponies? Really? What did we do before auto-correct?

  6. I find modus ponies entirely convincing. The rest I'm going to have to think about.

  7. I'll never be able to think of it as anything else but 'modus ponies' now. I find that oddly enriching

  8. If we only taught logic in kindergarten it would have entered the lexicon long ago.