Jump to ContentJump to Main Navigation
Making Things Up$

Karen Bennett

Print publication date: 2017

Print ISBN-13: 9780199682683

Published to Oxford Scholarship Online: June 2017

DOI: 10.1093/oso/9780199682683.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 20 November 2017

(p.239) Appendix: Objections to the Second Grade of Causal Involvement

(p.239) Appendix: Objections to the Second Grade of Causal Involvement

Making Things Up

Karen Bennett

Oxford University Press

I noted back in Chapter 4 that what I there called “the second grade of causal involvement” seems to prompt a lot of objections. Here are six. The first five are to the effect that I have illicitly assumed some controversial doctrine or other; the final one is to the effect that there is something wrong with Truth Conditions 2.

Objection 1:“You’ve assumed that ‘vertical’ composition occurs. What about compositional nihilism, à la van Inwagen (1990), Merricks (2001), Dorr (2005), and Sider (2013)?”

Well, yes, I did assume that composition occurs. But this really does not matter very much. Remember that although my examples in §4.3.2 were very composition-ish, my general point is intended to be broader than that. Those who do not believe in composition almost certainly believe in other vertical building relations. Consider the sort of claim often attributed to nihilists who wish to recapture ordinary talk by paraphrasing it in ways that do not quantify over composites1—for example, that although there are no tables, there indeed are simples-arranged-tablewise. But what is this property being arranged tablewise that the simples plurally instantiate? Whatever it is, it surely is not fundamental; simples plurally instantiate it in virtue of instantiating other properties, some physical, some functional.2 So the nihilist who wishes to recapture ordinary talk needs to say that there are building relations between properties. That is, your average compositional nihilist holds a view much weaker than true “flatworldism”, as I characterize it in Chapter 8—i.e., much weaker than the view that there are no building relations at all, and that nothing is more fundamental than anything else. At the time of writing, I know of no true flatworlders. And I see no reason to doubt that the basic gist of §4.3.2 can be recapitulated in terms of some other vertical building relation.

It is also worth noting that not even compositional nihilism itself is very widespread. Sider (2013) and Dorr (2005) do defend a full-blown version, according to which nothing is ever part of anything else. But Merricks and van Inwagen instead defend weaker views. van Inwagen believes in composite living organisms (1990), and Merricks believes in composite conscious beings (2001). These things are presumably put together by means of diagonal building, as per the second grade of causal involvement. Indeed, van Inwagen is fairly explicit that this is what happens when a sperm and an egg come together to make a zygote (1990, 151–3).

(p.240) Objection 2:“You’ve assumed eternalism. No presentist will buy this.”

I admit that much of the chapter is, for simplicity, written in eternalese—I have, for example, quantified over times in a way that is easiest to interpret given eternalism. However, but there is no stable dialectical position from which there is an objection here. Eternalists will obviously not care if accommodating diagonal building requires eternalism. Presentists will care, but they should not believe that it does. Presentists are often accused of having trouble accounting for cross-time relations, and they typically attempt to resist the accusation one way or another (e.g. Markosian 2004). Diagonal building is a cross-time relation, to be sure—that is one of the main points—but in that respect it is just like many others: admiring Lincoln, missing your dead mother, causation, motion. Whatever solution the presentist endorses in the other cases will be equally successful here. If there is no satisfactory solution, so much the worse for presentism.

Objection 3:“You’ve assumed that things come into and go out of existence. What about necessitism—the view that everything that exists necessarily exists (Linsky and Zalta (1994, 1996); Williamson (2002, 2013))?”

The idea here is this: if everything necessarily exists, then nothing is ever created or destroyed. If nothing is ever created or destroyed, then it is never the case that makings are destroyed as a composite comes to be. And if the makings are never destroyed, there are no cases of the sort that drove me from Truth Conditions 1 to Truth Conditions 2 of diagonal building; if composites never come to be at all, there is no diagonal building in the first place. Either way, there is no reason to take seriously a diachronic building-as-a-process relation distinct from (though analyzable in terms of) purely vertical building.

I myself do not believe necessitism; I share Robert Stalnaker’s (2012 contingentist inclinations (see my 2005, 2006)). But even assuming that necessitism is true, the objection fails. Here’s why.

Necessitists always try to do some justice to the intuition that some things exist only contingently. (In principle they need not do this, but in practice they always do.) So while they deny that anything exists contingently, necessitists instead say that there is a special property that some things have only contingently. Both Williamson and Linsky and Zalta take concreteness to be this special property. On their view, both the eggs and the cake exist necessarily, but are only contingently concrete. The eggs are not destroyed by the mixing and baking process, they just lose a lot of properties, and become nonconcrete. Similarly, the cake is not brought into being, but merely brought to concreteness.

So is our Williamsonian objector correct to say that, since everything always exists, nothing is diachronically built? No, and it should be clear why. Even if neither the cake itself nor the fact that it exists is built from the earlier ingredients, the fact that it is concrete, is a cake, and has various other properties is so built. By both Linsky and Zalta (1994, 446)3 and Williamson’s lights (2002, 245–6), nonconcrete entities have no nonmodal properties. The cake, however, has many, including being a cake. In short, although I did assume that some things are created and destroyed, the assumption is dispensable. (p.241) The arguments of §4.3.2 can be revised to reflect the view, which I myself do not hold, that everything exists necessarily.

Objection 4:“You’ve assumed the falsity of classical extensional mereology—in particular, the falsity of the axioms of unrestricted composition and extensionality.”

No, I didn’t. Regardless of whether or not those axioms really are false, §4.3.3 should have made clear that I would prefer that my arguments not turn on such controversial assumptions. So let me explain why it might seem as though my arguments assume the falsity of the axioms of unrestricted composition and extensionality, and then explain why they in fact do not.

The axioms in question are as follows. Unrestricted composition says that any things whatsoever have a fusion; composition is automatic. Extensionality, or uniqueness, says that no two things can have the very same parts—that x and y have the same parts just in case ‘they’ are identical. The reason it might appear that I have assumed that these axioms are false is that they appear to generate another route to the claim that no composites are ever created or destroyed. Like the challenge from Williamson, Linsky, and Zalta, this would mean that there is no pressure to acknowledge a diachronic building relation distinct from purely vertical building. (Unlike that challenge, it does not involve the claim that everything exists necessarily—nor even that all composites exist necessarily.)4

Here is why it might appear that uniqueness and unrestricted composition together entail that no composite object can be created or destroyed. Consider an egg that I use to make the cake. I break it and beat it and spread its parts around. But the parts still exist, so by unrestricted composition they still compose something. That is, there is a thing composed of exactly the same parts that once composed the egg. Since uniqueness says that sameness of parts is sufficient for identity, it follows that the egg is identical to that scattered thing. That renders my breaking rather ineffective; all I did was spread the egg out over a larger spatial area. So I didn’t destroy it in making the cake after all. Mutatis mutandis for my putative ‘making’ of the cake itself. Really, all I did was change the properties of and relations between the parts of a pre-existing fusion. I didn’t in fact create anything at all.

This line of thought is mistaken. Even if the axioms do entail that fusions cannot be created or destroyed—which is unclear at best, given that the axioms are silent about what it is for a fusion to persist over time—they do not entail that ordinary objects cannot be. That is because it is natural to deny that ordinary objects are fusions of spatial parts. One option is to claim that ordinary object terms like ‘egg’ refer to fusions of spatial and temporal parts. On this perdurantist line, the egg has temporal parts before it is broken and sufficiently mixed at t, but not after. It is not relevant that the spatial parts of the egg continue to exist and, by unrestricted composition, have a fusion after t; that fusion is not (p.242) a temporal part of the egg. (Mutatis mutandis for the cake; there can perfectly well exist a fusion of the spatial parts of the before the first temporal part of the cake itself.)

Another option, compatible with endurantism, is to claim that ordinary object terms refer to fusions that instantiate certain properties. If ordinary things of kind K are essentially F, then a fusion that once is a K ceases to be a K when it ceases to be F. Suppose for simplicity (this is clearly false) that eggs essentially have intact shells. Then the fusion that was the egg before the shell is cracked continues to exist when the shell is broken, but it ceases to be an egg. Similarly, in making the cake, I perhaps do not cause it to be the case that some parts compose some fusion, but I do cause it to be the case that that fusion instantiates the complex property of being a cake. On this line, the diagonal building relation would have to be analyzed in terms of the instantiation of nonfundamental properties rather than the composition of nonfundamental objects—but this matters not, as it is building either way.

Objection 5:“Truth Conditions 2 does not accurately state the truth conditions for diachronic building sentences. The right hand side is not sufficient[5] for the truth of any diachronic building sentence.

(p.243) Suppose that instead of baking a cake out of the ingredients I have put on the counter, I hurl them all around the kitchen, making an epic mess. Doesn’t the relationship between the ingredients and the mess meet Truth Conditions 2? The mess shares a lot of parts with the ingredients—arguably more than the intact cake does. Yet to say that the mess is built from or diagonally composed of the ingredients seems…odd. True, we do say ‘Karen made a mess’, but surely it isn’t exactly a thing that I made. Or suppose that I do bake a cake, but then rip it into chunks for fondue or trifle. The relation between the cake and the chunks—as well as the relationship between the ingredients and the chunks—also seems to meet Truth Conditions 2. But isn’t that relation decompositional? It is a matter of taking things apart, not putting things together.

My main reply here is very short: if Truth Conditions 2 is false, so much the better for me. I simply assumed that it is true, and argued that its truth provides no reason to deny that there are diachronic building relations. If it is not true, well, then it certainly provides no reason to deny that there are diachronic building relations! And matters look even better for diachronic building if it further turns out that mild tweaking doesn’t help, that it is simply not possible to provide truth-conditions for diachronic building sentences that do not mention diachronic building. In that case, the motivation to take diachronic building as nonfundamental disappears. So worries about Truth Conditions 2 are all in my favor.

Still, I am inclined to think Truth Conditions 2 is true, and that the issues raised by this objection are independently interesting. I thus step aside from the main dialectic of the chapter to explore this further.

Could Truth Conditions 2 be refined to avoid classifying activities like making messes and piles of crumbs as building processes? One could try. A natural option would be to insist further upon a feature that is in fact already part of Truth Conditions 2 as formulated: that diachronic building be many–one. This would block the crumbs case, and arguably the mess case.6 But this move seems to miss the point. For one thing, I explicitly allowed disunifying building back in Chapter 3. For another thing, there are other examples that are many–one, but challenge the sufficiency of Truth Conditions 2 in related ways. Suppose I make compost from a pail of vegetable scraps, take it out to the garden, and some ‘volunteer’ tomato plants grow from seeds that were in the compost. This case meets Truth Conditions 2 even holding fixed its many–one structure: a lot of the parts of the scraps, at some level of decomposition, persist as parts of the (single) tomato plant. But it still sounds odd to say that the tomato plant is made from or built out of the vegetable scraps.

An alternative way to modify Truth Conditions 2 would be to include some constraint on how much time can pass between t1 and t2, or perhaps better, on how much macrolevel qualitative change can occur between t1 and t2. Or perhaps the purposely vague “most of the zzs are among the xxs, and vice versa” clause could be precisified. There are presumably other moves available.

Still, though, this project strikes me as hopeless. We don’t have firm enough judgments about every single case to be sure exactly which ones to rule in and which to rule out; an (p.244) extensionally adequate analysis is not a meaningful possibility here. And revising Truth Conditions 2 even in just the ways above would yield artificial distinctions. Consider, for example, the idea that somehow the tomato plant case must be ruled out because there is too long a time lapse, or too much qualitative change, since the pail of scraps in the kitchen. That is nonsense. There is no real difference in kind between the amount of time and qualitative change involved in the change from scraps to tomato plant and the amount of time and qualitative change involved when grapes are made into wine—or, even further down the road, into cognac or vinegar. Not all of our initial intuitions here are to be trusted,7 and we should take care to avoid endless and pointless ad-hoc refinements.

So what is my attitude towards the putative counterexamples to Truth Conditions 2? It is this: let them go. We should not take the ‘troublesome’ cases as clear counterexamples to the sufficiency of Truth Conditions 2, and we should not scramble to revise and repair in light of them. The proper response, instead, is to take them at face value, as somewhat fringe examples of the concept under investigation. It is a mistake to insist that there are hard and fast lines between the ‘troublesome’ cases raised here and the ones I have been treating as central examples of diagonal building.

Further, insofar as the troublesome cases are troublesome, it has little to do with the diagonality or building and everything to do with the mereologicalness of the cases. That is, it matters that I have, for clarity, focused on examples that centrally involve composition. What bothers us here are a host of interrelated questions about what counts as creating an entity, rather than destroying one, or merely rearranging some stuff. Addressing such issues requires answering the Special Composition Question, and telling us what counts as a single composite object. But notice what remains, even leaving such questions alone. It is indisputable that all of these cases—from making a Lego® castle to making a cake to making a mess or growing a tomato plant—involve rearranging microlevel entities into new configurations. That is why they meet the basic shape of Truth Conditions 2. But what is it to rearrange microlevel entities into new configurations, changing what macrolevel entities exist, and what macrolevel properties are instantiated? It is to bring about various changes in what builds what. The ‘macrolevel’ is the built. Thus all of these cases—all of them—involve diagonal, diachronic building. Perhaps they do not all involve diagonal composition, as captured in the letter of Truth Conditions 2. But they all involve diagonal building, perhaps versions involving grounding, realization, or microbasing. Any talk of changing the macrolevel by changing the microlevel is talk of diagonal building.

The fact is that these cases—the castle, cake, mess, and tomato plant—shade into each other. They all involve the rearrangement of underlying smaller bits; they all involve matter that composes-or-grounds-or-realizes-or-microbases something coming to compose-or-ground-or-realize-or-microbase something else. As soon as we get anywhere near an adequate answer to the question of what diagonal building is, we cannot help but (p.245) slide into what might seem to be a more general idea—something like things turning into other things, becoming other things. The scraps eventually turn into the plants, just as the eggs and flour and so forth first become batter, then cake, then crumbs. The only way to stop this slide into general becoming is to stop quite early on, at my move from Truth Conditions 1 to Truth Conditions 2, and insist that making a cake does not involve the same relation(s) as making a castle out of Legos®.

In short, there is a continuum, ranging from straight-up, synchronic/vertical composition at one extreme, through the diachronized but still essentially vertical version that is captured in Truth Conditions 1, through the truly diagonal version captured by Truth Conditions 2, through to various other forms of becoming, change, and decomposition at the other extreme. This is no problem for my overall claims about the second grade of causal involvement, nor for my discussion of the second kind of causal taint more generally. Once we start thinking diachronically, about building as a process, the notion of building gets tangled up with notions of causation and persistence. (p.246)


(1) Difference-minimizing nihilists who wish to up-play their expressive power, as I put it in my 2009.

(2) I have argued elsewhere that compositional nihilists who wish to recapture ordinary talk are committed to countenancing properties which non-nihilists need not countenance (2009, 64). Here, however, I am not relying on that strong claim.

(3) Linsky and Zalta’s view is more subtle than this, but the added complexity does not affect my basic point.

(4) Even assuming (a) the truth of unrestricted composition, (b) uniqueness, and (c) that they together entail that no composites are ever created or destroyed (which I deny in the main text), it still would not follow that all composites exist necessarily. If it is contingent what simples exist, it is contingent what fusions exist. Further, Ross Cameron has argued that the principle of unrestricted composition is itself contingent (2007)—on that view, it is contingent what composites exist, even holding fixed what simples exist, and assuming the actual truth of unrestricted composition.

(5) Two possible objections to the necessity of Truth Conditions 2 are perhaps worth mentioning, though only in a footnote.

First, suppose one thought that the identity and existence of parts depended upon the identity and existence of the thing of which they are parts. (Aristotle, for example, thought that nothing could be an eye or a kidney unless it was part of a functioning human body. The view in question here goes further. It is not just that a certain lump of flesh ceases to be a kidney when removed from the body to which it belongs, or when that body is destroyed; rather, that lump of flesh ceases to exist altogether, and is replaced by a numerically distinct object.) On such a view, ordinary cakes are not diachronically composed of their ingredients. No parts of the eggs survive to become parts of the cake, and so by Truth Conditions 2 the eggs do not diachronically compose the cake. (Thanks to Ross Cameron here.)

I am not very concerned about this objection. The letter of Truth Conditions 2 does presuppose that such a view is incorrect, which I think is not a crazy presupposition. Besides, the spirit of Truth Conditions 2 survives. Anyone wishing to endorse the above view simply needs to modify it slightly. The eggs still diachronically compose the cake in virtue of underlying causal processes; the cake does not come to be exnihilo. It is just that on this view, those underlying causal processes do not involve persistence.

The second objection to the necessity of Truth Conditions 2 can be seen by reflecting upon a version of the ship of Theseus case. Suppose it was originally composed of pine planks, which have gradually been replaced over time so that the ship is now composed of pieces of aluminum, and has no matter in common with the pine planks of which it was originally built. Is the ship made from those pine planks? If it is, then Truth Conditions 2 is not necessary for the sort of diagonal composition in question. (If not, there is no puzzle here.) I myself occasionally feel a bit of pressure to say that Theseus’ ship is made from the pine planks, though I do not really endorse this intuition, and find that it is not widely shared.

If that intuition is correct, though, there is a relatively simple fix. Accommodating Theseus-style gradual replacement of the things that compose the xxs, just requires moving to a two-stage definition that allows for stepwise chains. First, call Truth Conditions 2 a definition of direct diachronic r diagonal composition. Second, say that the xxs stand in the diachronic composition relation to y iff

Either the xxs directly diachronically compose y, or

There is a stepwise chain of direct diachronic composition between the xxs and y.

That will handle the ship of Theseus case.

(6) Ruling out the mess example on the grounds that it isn’t many–one would require denying that the mess is a single entity, which in turn would having an account of what a single entity is.

(7) We should not, for example, expect the fact that we say that wine is made from grapes, but speak of ‘wine vinegar’ rather than ‘grape vinegar’ (and ‘cider vinegar’ rather than ‘apple vinegar’), to be reflected in a real metaphysical difference between the relation between these grapes and that bottle of wine and the relation between those grapes and that bottle of vinegar. There is no relevant difference between the cases.