Metametaphysics: Review, Commentary, and Discussion (Part 1)

For those who may be paying attention to my recent posts, I am currently reading the collection of essays Metametaphysics, which talks about how metaphysics ought to be done. There is a lot of discussion about whether problems in ontology, such as mereological sums (if there is a tablewise arrangement of atoms, does some “new” object that we call a table come into existence, or is that just a shorthand way we talk about such tablewise arrangements of atoms?), are just semantic. In other words, when I say that a table is nothing more than a tablewise arrangement of atoms, and you say that a table is something above and beyond the tablewise arrangement of atoms, are we simply just using the word “table” in different ways, thus resulting in the differences in how we conceptualize what a table is? Here I am going to discuss (more so than review) the first three essays in this collection.

Metametaphysics, edited by David J. Chalmers, David Manley, and Ryan Wasserman, Copyright 2009, Oxford University Press, 540 pages

Essay 1: “Composition, Colocations, and Metaontology” by Karen Bennett

Essay 2: “Ontological Anti-Realism” by David J. Chalmers

Essay 3: “Carnap and Ontological Pluralism” by Matti Eklund

“Composition, Colocations, and Metaontology” by Karen Bennett

Karen proposes that realism and nihilism of composition (mereological sums) have the same problems:

1) when, if ever, is it true that tableness exists? When, if ever, is it true that atoms are arranged tablewise?

2) if there is a table X composed of molecules (m1 + m2 + m3 + … + mn) = sum(mn) then taking away mn will also result in there being a table, so is there a table X(sum(mn)) and a table X(sum(mn-1)) that co-locate in the same region? This applies to both the existence of tableness and the property of being arranged tablewise.

My own view (on problems 1 and 2) is that what matters when discerning whether a table is a table is the information that the concept of table gives about the object: does applying the concept of tableness to the tablewise arrangement of atoms help us make predictions (reduce uncertainty) about the tablewise arrangement of atoms? The application of the concept of tableness, though grounded in the information that this concept brings to bear, can be somewhat arbitrary: a 3-foot tall boulder can be used in the same way as a table (i.e. I can make similar reliable predictions about how I arrange atoms relative to the boulder as I can about how I arrange atoms relative to a table). The tableness of something, then can be context dependent.

I expound on these ideas in much greater detail here.

3) overdetermination: saying that tableness exists does not explain anything more than saying atoms arranged tablewise does: the tableness of the atoms arranged tablewise does not explain why the table can be used as a table (e.g. why I can set my cup down on it); this can still be explained by the atoms being arranged tablewise.

Bennett says that as long as someone claims that composition (tableness) is necessary whenever atoms are arranged tablewise (i.e. conditionals linking the simples to the existence of a composite are necessarily true), then tableness is true. Thus, there is no overdetermination problem: it just is necessary that when atoms are arranged tablewise, the result is tableness.

For me, I would say that small changes in the arrangement of atoms tablewise may have small effects on the tableness of the object, but they are still changes in the information of the table – they will change some of the predictions you can make about how the tablewise arrangement of atoms will interact with any other arrangement of atoms

4) both realism and nihilism require that multiple objects (identified by their persistence conditions) can co-locate in a region of space: a statue and the a lump (the realist) and the arrangement of atoms that-would-survive-squashing-wise vs the arrangement of atoms that-would-not-survive-squashing-wise (the nihilist).

Once again, I refer to information: what predictions (reductions of uncertainty) can I make about the lump of clay when it is considered as just a lump of clay and what predictions (reductions of uncertainty) can I make about the lump of clay when it is considered a statue?

This same sort of colocation could work on things that are less distinct than a lump of clay and a statue. For instance, a tablewise arrangement of atoms could be seen as two objects thusly: a table with a cup and a plate set on it at locations x and y respectively is, at the same time, a tablewise arrangement of atoms onto which a cupwise arrangement of atoms is set at location x and a tablewise arrangement of atoms onto which a platewise arrangement of atoms is set at location y. Do these two different things mean that the table is two different objects, given that it has a different relationship between the two objects?

Once again, I don’t think it’s necessary to imbue a different tableness of the table given these two relationships just as it is not necessary to give a different thingness to a lump of clay formed into a statue. In both cases, what we’re interested in is the information: what does a particular relationship (the table to the two objects; the arrangement of atoms in the clay statue) allow us to predict (or retrodict)? At the ontologically significant level, the information changes when objects are set on a table or when clay is formed into a statue.

In the case of the table-cup-plate system, when we are evaluating tableness, we have to ask: is the information changed enough that the concept of tableness no longer gives accurate information (i.e. we cannot make accurate predictions about how the table-cup-plate system will behave given changes in arrangements of atoms outside the table-cup-plate system)? This obviously depends on relevant information, such as the size and sturdiness of the table, the plate, and the cup; whether the plate and cup can be moved around the table; and so on. If we can still make predictions (reductions in uncertainty) about the table-cup-plate system that still allows the system to satisfy the criteria for tableness, then it is still a table. However, for instance, if so many objects were set on the table such that the table-objects system turned into a heap of objects, then the concept of heapness would yield better predictions about the system.

As for the clay statue and whether it has statueness or lumpness, the application of each concept becomes context specific: I will apply the concept of statueness when that allows me to make certain predictions accurately and I will apply the concept of lumpness when that allows me to make certain predictions accurately. The alternations between statueness and lumpness would be sort of like the vase vs two faces illusion – they both are instantiated, but the conceptual application can alternate.


“Ontological Anti-Realism” by David J. Chalmers

Chalmers says that there is a difference between internal (everyday conversational) and external (ontological) claims: Carnapian internal propositions are more like ordinary (everyday) notions (“tables exist” is trivial since we all know tables are something we use) and external propositions are ontological notions (“tables exist” is non-trivial because it depends on whether mereological sums actually exist or not).

This means, in the ordinary claim, even if it is the case that tableness does not exist (there are no mereological sums), it is still correct, even if not true, to say that “this is a table”: even if all that is actually true is that a tablewise arrangement of atoms exists (no tableness above and beyond the tablewise arrangement of atoms), it is still correct, in the internally coherent language of everyday conversation, that there is a table. Ordinary claims and ontological claims have different correctness conditions – different criteria for evaluating whether a statement (e.g. this is a table) is correct: ordinary correctness conditions would take the statement “this is a table” to be correct if there is something present that we call a table; the ontologist would take the statement “this is a table” to be correct if it is true that tables actually exist. Thus, there is potentially a difference between a proposition being true and being correct. Which means, using the different internal claims for different schools of thought, we get:

Nihilist: “there is a table” is correct if and only if “there are particles arranged tablewise” is true – conditionally correct

Nominalist: “there are prime numbers” is correct if and only if it is true according to the fiction of mathematics that there are prime numbers – conditionally correct

Universalist: “there are two objects on the table” (as opposed to three: the two simples and the mereological sum of the two simples) is true if and only if there are two familiar macroscopic objects on the table – conditionally correct

Chalmers wants to point out, though, that these are different models of reality rather than worlds in which different ontologies are true. The main difference between a model and a world is that a model comes with a built-in domain: all of the things that exist with respect to the model (a universalist model will include mereological sums in its domain while a nihilist model will not). A world, like our real, actual world, doesn’t (necessarily) come with a built-in domain, since the domain of what exists is what is in question: does the domain of everything that exists contain mereological sums (e.g. tableness) or not?

My commentary:

These models, I think, are an interesting way of conceiving of the information given by objects. A model is, in a sense, a tool for making predictions (attempting to reduce uncertainty). If we model the world as having mereological sums, such as tableness, does that model allow us to make better predictions about how objects will behave in different circumstances? In other words, if I model tableness onto tablewise arrangements of atoms, does that give me information?

I would say that it does, in the sense that tableness is a concept in our minds. Applying the tableness concept is synonymous with using a model in which tableness is in the domain of existing things, and this model will give me more accurate predictions than if I treat the tablewise arrangement of atoms as if there is nothing above and beyond the tablewise arrangement of atoms.

“Carnap and Ontological Pluralism” by Matti Eklund

Eklund discusses how Carnap made the distinction between propositions internal to some framework and propositions external to some framework; Eklund further breaks external down to pragmatic-external (deciding which framework is best to use) and factual-external (trying to decide if what is said internal to a framework applies to ideas outside the framework, such as to actual ontological objects).

A particular framework “A” would be something like a language (perhaps a version of English, or a way of understanding English), where a proposition such as “tomatoes are a fruit” is true, whereas a particular framework “B” would be a different language where the proposition “tomatoes are a fruit” is false.

An issue often levied at Carnap is ontological pluralism: framework A is just as valid as framework B, they are simply employed for different uses or different contexts (or by different people and/or cultures). Some go further and accuse this pluralism of relativism: if framework A says that female genital mutilation is bad and framework B says that female genital mutilation is good, then all we can say is that FGM is bad for those who subscribe to A but that it is good for those who subscribe to B, but we cannot make the external proposition that “FGM is bad” as any sort of universal or factual statement.

I am not going to argue whether or not Carnap was an ontological pluralist, though I don’t think he was (and neither does Eklund). I will say, though, that I also don’t think pluralism is true. If there are two frameworks, Fa and Fb, then a person Pa who subscribes to Fa will say that Fb is untrue because Fb is not true in Fa. That means any person Pb who subscribes to Fb, when speaking using the Fb framework, will not be saying anything true in Fa, and therefore Pa will have to consider Pb wrong. If some proposition M in is true in Fa and also true in Fb, in what way can we say that it is true in both Fa and Fb, since neither considers the other correct? Given that the frameworks Fa and Fb take different facts to be fundamental, the truth of M would be justified differently in Fa then in Fb. For instance, if we both take it to be true that the universe came into existence 13.8 billion years ago, but within my framework I take it to be true that it came into existence without the will and actions of a creator but you take it to be true that it came into existence because of the will and actions of a creator, then when we both say “the universe came into existence 13.8 billion years ago” are we really saying the same thing? On the face of it, yes, and lets even say that it is the case that the universe came into existence 13.8 billion years ago: both of us (and our associated frameworks) are saying something true, but are both of us (and our associated frameworks) right? One of us is incorrect about how the universe came into existence 13.8 billion years ago, and therefore their framework is unjustified in its true proposition that the universe came into existence 13.8 billion years ago – the conclusion about the age of the universe is grounded in something that is not the case. In this sense, both of us are compelled to think that the other, given our own frameworks, is wrong. This means that we cannot take a pluralistic view of Carnapian frameworks.

However, my main goal here is to adapt this idea of frameworks into the concept of paradigms. In science, a paradigm is a set of laws and/or theories that are used as if they are true, even if we have evidence that they are not true, because they can still be used to make good predictions in most cases. The canonical example of this is Newtonian gravity: we know it’s not actually true, because we have Einstein’s theory of general relativity, but Newtonian gravity can still be used in most cases because it gives us true predictions. This is a good example because here we already have a new paradigm – general relativity – that supersedes the previous paradigm, showing how something we know isn’t true (since we have the truth, or at least a better approximation of the truth) can still make useful predictions. It is not, however, necessary that we already have some better paradigm: for instance, we use the standard model of particle physics, which makes great predictions, even though we know there are things wrong with it (like that it doesn’t account for gravity). Once new theories with new experimental evidence come in, a new paradigm that supersedes the standard model will be adopted, but until then, the standard model can still be used to make accurate predictions.

I would liken a Carnapian framework to a paradigm: we use a particular framework (language, understanding of a language, or to be very general, a philosophical system like the scientific method or classical liberalism or realism) because using that framework allows us to make accurate predictions. In the case of the scientific method, the framework has been extremely successful (within the realm of scientific inquiry); with something like classical liberalism, it has been successful, but there are definite flaws (certain aspects of human nature lead to its failure); realism is successful because taking mereological sums to be true (if there is a tablewise arrangements of atoms, then there is a mereological sum above and beyond the atomic makeup that is a table) is useful for practical reasons (instead of saying “set this cupwise arrangement of atoms on that tablewise arrangement of atoms” I can say “set this cup on that table”) and for more trivial reasons (when I go to buy a table, the salesperson doesn’t ask if I mean to buy every atom arranged tablewise or just some of them).

The important part, though, is that we are still using information: a framework, or paradigm, is useful because it is capable of conveying necessary and sufficient information; we can make predictions (reduce uncertainty) better using some frameworks better than we can with others – Newtonian gravity makes accurate predictions in most cases, but not all cases, making it a less suitable paradigm than general relativity.

This paradigm view of frameworks makes internal propositions that are analytically true and external propositions that are conditional in the form of predictions. For instance, in Newtonian physics, it is analytically true that “gravity is the attractive force between two objects determined by the product of the gravitational constant multiplied by the two masses, all divided by the square of the distance between their centers of gravity”. An external proposition in the paradigm framework of Newtonian physics would be something like “if I throw object x straight upward from sea level on planet earth with force F, then I predict that it will go up h meters before falling back to earth and hitting the ground at velocity v”. This is an external proposition because, although we represent objects within the paradigm framework of Newtonian physics – distances, forces, velocities, and so forth – we are actually making predictions about objects in the real world that can then be tested, verified, falsified, etc, which will then either confirm or infirm the paradigm framework. For instance, if you use Newtonian gravity to make a prediction about how throwing an object at 99.999999999% the speed of light away from the event horizon of a black hole will play out, you will infirm the Newtonian paradigm framework of gravity – the actual world has shown the paradigm framework to be deficient.

This would also go for paradigm frameworks that are impoverished as well – paradigm frameworks that leave out or refuse to acknowledge aspects of a paradigm framework that make reliable predictions; or paradigm frameworks that make unreliable predictions in significant realms of relevant inquiry. A good example of this would be intelligent design, which has little to no predictive power (it would require a field of psychology for the designer rather than a biological framework to even get started), it refuses to acknowledge inconvenient evidence (like molecular and anatomical homology), and it posits mechanisms that don’t allow for testable hypotheses (there is no way to gain information – no way to reduce uncertainty).

This concludes my review, commentary, and discussion of the first three essays in Metametaphysics. I guess it would be more accurate to say this is less a review, commentary, and discussion of the first three essays and more of a discussion prompted by the first three essays. I’ll do more as I continue reading through the collection of essays.