Friday, October 17, 2008

Schaffer on truthmakers and ontological commitment

In a few places, I've argued that Quine was wrong about ontological commitment. If you want to know what a sentence is ontologically committing to, look not to what the quantifier must range over, say I, but to what must exist to make it true.

Jonathan Schaffer replied in his 'Truthmaker Commitments'. He argues that the motivations for my view are bad ones and goes on to offer some objections to it. He does argue that truthmakers play a role though: but it's not in identifying the ontological commitments, but in identifying what is fundamental according to the theory.

I've just written my reply. I argue that the motivations are good ones, and I aim to counter the objections. I hope the view becomes a bit clearer in my responses to the objections - certainly, they forced me to say some things I hadn't said in the paper Schaffer is criticising. I end, though, by suggesting that it's not obvious there's a genuine dispute between me and Schaffer - that we just mean something different by 'ontological commitment'.

Comments, of course, would be welcome.
Update: the paper has been revised as of 18/10/08


N. Wildman said...

I was wondering how your story is really all that different from the Quinean's. The notion of ontological commitment that you support is quantification (of the 'most natural quantifier'), so OC and quantification come together, but not in English. But Quine's story also claims that English quantification isn't a guide to OC, but instead first order formal quantification is. What's the difference? If it is just that you don't accept first order formal quantification as the 'most natural', then it doesn't look like you should give up Quinean OC. Instead, you should just re-formulate it in terms of this new, most natural, quantifier and the language which expresses it. This sort of re-formulation is totally consistent with the Quinean method - so how are you giving it up?

Also, does your turning to some 'most natural quantifier' damage your taking the English sentences to be literally true? I would think that taking an English claim like "A table exists" as literally true means accepting that it is making an ontological claim about the existence of a table. But on your reading, it isn't - not really, anyway. So what do we mean by 'literally true', or what do you think the claim is expressing?

Sorry if I'm way off, or if the questions don't make much sense.

Ross Cameron said...

I try to address your first point in the section in the paper 'Quine's revenge'. It's true that I'm a Quinean about the language that uses only perfectly terms (call it 'Ontologese'). The ontology of the world is what can be said to exist, when 'exists' is the most natural quantifier, so what there *really is* is what must be amongst the values of the variables bound by that quantifier.

But there are significant differences between myself and the Quinean. I don't demand that to avoid commitment to Fs you either have to give up F-talk or paraphrase it away by providing a translation to sentences in English that play the same theoretical role that don't quantify over Fs. I don't demand that Ontologese be a fragment of natural language: it may not be speakable or graspable by us.

Also, I think that there are sentences that are trivially true, that involve quantification over the Fs but which don't have any ontological commitments (and so don't correspond to any sentence in ontologese). The truths of mathematics are amongst these: I think 'there are numbers' is literally true but that its truth just doesn't demand anything of the world, and so doesn't commit us to anything, a fortiori not to numbers. And I think the story can largely stop there - there's simply no need to 'paraphrase' away the mathematical claims to avoid commitment to numbers.

So both those I see as big differences. How does that strike you?

I don't really get your second question, I'm afraid - maybe you can elaborate. By 'literally true' I just mean true, not true according to some fiction, or quasi-true or what have you - just, true. I do think 'A table exists' makes an ontological claim about the existence of a table, and I think it's literally true. I think there are tables. I just don't think that ontological claim perspicuously reveals the ontological structure of the world, because it's expressed using an unnatural quantifier.

Compare the following. 'These things are grue' might be literally true. And it makes a claim about the properties things have. So these things *are* grue. But I don't think that claim perspiuously reveals the structure of reality either, because it's expressed using an unnatural predicate. Better to say, e.g., 'these things are green, and the time is t'. Likewise, better to say 'there *really are* some simples arranged tablewise'. Does that help at all?

Anonymous said...

I like that paper and I think it does defuse most of Schaffer’s criticisms. However, I thought you might use the last section a bit more to attack the bits of Schaffer’s view that distinguishe it from yours. Here are 3 (not too original) ideas.


“I know why I say we shouldn’t count the derivative entities: it’s because there aren’t really any such things – it’s merely a matter of convention that we can truly describe the world thus. But if Schaffer insists that the derivative should be taken with more ontological weight, I think he owes us an explanation for why they don’t count against a theory when weighing up its ontological costs.”

I totally agree; this is what puzzles me about Schaffer’s view most. His answer will be that if you stick a grounding-relation between a and b, one of them automatically becomes a free lunch. But how, if 'to exist' has only one meaning, could it not be inconsistent to claim
(1)a exists, b exists
(2)a ≠ b
(3)given a, the existence of b is
no addition to being (i.e., to
what exists)
How can it help to stick a relation – any relation? – between the distinct objects a and b to rid b of its ontological weight? Schaffer says that grounding relations are relations of abstraction but is largely silent on what abstraction is. So, I think the pressure is on him to give further explanation.


Here is another line one might push against Schaffer: If he wants to avoid Neo-Carnapianism about the derivative (“if what derivative entities there are is as much an objective feature of the world as what fundamental entities there are”), he will require a form of ontological maximalism (e.g. universalism as far as composition is concerned). For, (1) why should grounding relations mirror just our common sense ontology (and not, say, the common sense ontology of other possible communities)? And (2) surely grounding relations should be more systematic than, say, our common sense theory of (de-)composition; maximalism would grant such a systematicity.
However, maximalism flies in the face of commons sense (e.g., There is then no way to truly say “There is no object that is the sum of you and Edgecliffe” with an unrestricted quantifier. I know you are sympathetic to the restricted-quantifier move, but surely one doesn't have to be such sympathetic:). This is bad, because permissivism is one of Schaffer’s two main motivations for Neo-Aristotelianism – and permissivism is probably motivated by reference to Moorean common sense. It thus seems Neo-Aristotelianism undermines its own motivation.
On the other hand, if grounding-relations were relative to conceptual schemes (Schaffer allows for this possibility at some places), they resemble truth-making relations a lot! You might try to argue that in this case Neo-Aristotelianism collapses into your view.


Another point that could work against Schaffer is the modal weirdness of his grounding relations. What explains the necessary connection between what's fundamental and grounding relations? Why couldn't there be just the fundamental reality as it is, without any derivative entities? When I understand your theory correctly, sentences such as “If there are particles arrangeD tabLEWISe, there is a table” is necessary, because they are, in a sense, analytic – implicit stipulations that govern our use of the (English) quantifier. On Schaffer’s view they reflect brute laws of metaphysics ('brute' in the sense that the underlying grounding relation is not further explained) ... this should raise suspicion.


Ross Cameron said...

Thanks Tobias. I agree completely on points (1) and (3). Could you say a little more about (2), though? I wasn't quite following you. I agree Schaffer has to go maximalist, but wasn't seeing the rest of it. Maybe if you could spell out what you mean by 'neo-Aristotelianism' and 'permissivism', and why you take the latter to be motivated by Moorean common sense - that'd be helpful.

Anonymous said...

Hi Ross

In his “On what Grounds what”, Schaffer gives his main motivations for the view that ontology is about discovering grounding relations between objects and discovering the fundamental (rather than discovering what there is). One of them is "permissivism", according to which Quinean existence questions are trivial – all the allegedly controversial questions about the existence of numbers, properties, fictional entities, etc. are trivially answered affirmatively by common sense: all these things do exist.
Once we accept that not ontology but common sense answers the existence questions and once we accept that there is an abundance of entities, it seems plausible to view the task of ontology as accounting for this abundance by means of a sparse basis; and this is what Schaffer's grounding-story (Neo-Aristotelianism) is supposed to do.

In order to establish permissivism, Schaffer uses arguments of the form:

1.There are prime numbers
2.Hence, there are numbers

1.My body has proper parts
2.Hence, there are composites

1.There are properties that you and I share
2.Hence, there are properties

1.Doyle created Sherlock Holmes
2.Hence, Sherlock Holmes exists

What we have here as premises are, according to Schaffer, “a mathematical truism”, “a biological triviality”, “an everyday truism”, “a literary fact” - all of which are more credible than any philosophical insight could be. (He dubs the mathematical premise a “Moorean certainty” - but it seems he would extend this to the other premises as well.) Thus, all sorts of things the existence of which is regarded as controversial by the Quinean do exist.

Now, Consider:

(*) “There is no object that has only the sun and your laptop as its parts!”

If (ontological) maximalism is true, then (*) is false (if the quantifier in (*) is unrestricted). However, (*) seems to be as much a Moorean certainty as any of the premises above in the arguments which, taken together, are meant to establish permissivism. Thus, by parity of reasoning, Schaffer should not only accept “There are prime numbers” as a premise in an argument for permissivism, but also (*) as premise in an argument against maximalism. Given that maximalism seems to 'follow' from Schaffer's demands on grounding, he cannot use Moorean certainties in an argument for permissivism.

(( I think that is a crucial difference to truthmaker theory which can (in principle) account not only for the truth of our beliefs about what there is but also for the truths of our beliefs about what there is not. This enables the truthmaker theorist to use common sense to depart from the Quinean view (i.e. truthmaker theorists can use Moorean certainties or maybe a principle of charity to establish that common sense existence assertions are largely true and take it from there; Schaffer can't). ))

As it stands, the worry is that one of the motivations for introducing grounding relations and for declaring them to be the proper subject matter of ontology is not well argued on Schaffer's view. Surely, Schaffer can still rely on other motivations for his view (he can, e.g., argue that we need a notion of dependence to raise particular ontological questions). However, permissivism might be more than just a motivation for a departure from the Quinean view (from the 'what exists?' to the 'how does it exist?'). One can read Schaffer as providing a permissivism-based argument for the view that Occam's razor should only be applied to the fundamental: Schaffer's alternative to a (traditionally understood) Occamian razor is the “bang for the buck principle”, meaning that the best theory is as sparse as possible at the basis and as abundant as possible on the derivative side. If permissivism is true and if there is only one sense of “to exist”, then Occam's razor is false (or not a good principle after all). In this case one might think that the bang for the buck principle is the only plausible methodological principle in the vicinity, which maintains at least the spirit of the razor.
Note that this motivation for the 'bang for the buck principle' does not rely on an independent understanding of grounding as providing free lunches. Rather, once we accept the need for the new principle, we should accept that we must be able to make sense of grounding as delivering free lunches somehow.
However, the worry is that Schaffer cannot appeal to Moorean certainties to support permissivism. But what other support does he have for the view? Again, if he does not provide independent support, then Occam's razor seems to trump permissivism and we do not need to buy into a rival methodological principle. So this argumentative route seems blocked.

Hope that is a little clearer.

Sanborn said...


I'm interested in reading your reply to Schaffer but the above link is broken. Would you be willing to post an updated link?