First Gunk, now Junk: Infinite Chains of Metaphysical Explanation

I recently read a few philosophy papers which share a common theme. They advocate the idea that there may be no foundational or basic metaphysical level of reality (whether monistic or pluralistic), and that therefore one should (or at least can) embrace infinite chains of metaphysical explanation.

Einar Duenger Bohn, in the draft paper “Must There Be a Top Level?” notes that many philosophers have discussed the conceivability of “gunk”, which is infinitely divisible stuff (a world is gunky if everything in it has, mereologically speaking, proper parts). Bohn thinks that “junk” is also conceivable, where everything is a proper part of something else.

The term junk is drawn from the paper “Monism: The Priority of the Whole” by Jonathan Schaffer (here’s a post which discussed an earlier draft of this paper). Schaffer argued against the conceivability of junk as part of a larger set of arguments in favor of the monistic whole as the foundational entity in a world. (To fill out the glossary, Bohn introduces the adjective “hunky” to describe a world both gunky and junky).

Schaffer says that in discussing possible worlds, a “junky world” makes no sense, since “world” refers to a single entity. Junk has no upward “cut-off” point to contain it within a world. Bohn objects that there is no need to constrain the term world as singular: maybe it can refer to a set or some other plural entity, which allows for junk.

But is junk really conceivable? Bohn offers thought experiments which put junk on a par with gunk with regard to conceivability (imagine each atom in a world is itself a world: now take this idea both upward and downward). What’s wrong with a “hunky” world extending without foundation in both directions?

Schaffer’s paper is also the target of Matteo Morganti’s just published “Ontological Priority, Fundamentality and Monism” (hat tip; no draft available online that I can see). Morganti sees no compelling argument for a foundational level of reality vs. the alternative of “metaphysical infinitism”. He also takes issue with Ross Cameron’s conclusion that we should postulate a fundamental level for methodological reasons if we are to have any satisfying metaphysical explanations (I discussed the relevant Cameron paper in this post).

A third paper, by Francesco Orilia, makes an analogous argument in a different philosophical argument “thread”: he compares whether facts (or states of affairs), which bind an object with its attribute (or relate multiple objects), are basic ontological entities (as in Armstrong), or whether they give rise to an infinite regress of binding relations (Bradley’s regress). Orilia thinks we should accept “fact infinitism” as a live option. (For a very deep dive on this see Bill Vallicella's thoughts and dialogue with Orilia here and here).

So is there a foundational entity or entities or not? Do you really have an explanation when you invoke an infinite chain? There’s a lot to digest here, and I plan on rereading this set of papers and others. My intuition has always been that there must be a fundamental level of reality, but an argument is needed here, not an intuition. I think we can reject gunky/junky worlds. I will go ahead and sketch my speculative argument below (a previous gunk-inspired post with a good comment thread is here).

First, though, let me note that I’m very happy to continue to find members of a new generation of philosophers who are taking up “meaty” metaphysics as a primary focus. Philosophers like Schaffer and Cameron are doing serious and innovative metaphysics, and are provoking responses from other young philosophers (definition of young = anyone younger than me): great stuff.


My tentative position is this (what follows assumes modal realism). When we employ our (remarkable) ability to conceive of infinities we are in danger of making a particular error. As background, I endorse conceivability as a guide to possibility, and specifically think that our concept of what is logically possible maps to what is metaphysically possible. I think this total space of metaphysical possibilities is infinite and this is what grounds our concept of the infinite. Now, philosophers usually refer to this space of possibilities specifically as the space of possible worlds (I think this is rooted in methodological usefulness). However, our attempt to conceive of what is possible in a particular world may be actually importing content which may not fit in one world. We need to be careful about how we define a world.

Consider the actual world. Following David Lewis in one respect (but not in others), I think the best sense of “actual” to use is as an indexical. If I take the actual world to be the causally connected region centered on my point of view, then, based on a posteriori reasons, the actual world should arguably be considered finite: quantum physics as applied to space-time implies no infinite divisibility, and causal connections only extend so far. If all possible worlds are likewise defined as centered causally connected patches, I hypothesize that all “worlds” should be considered likewise finite.

Let me pause here for a moment. Many (most?) thinkers assume that a world is an isolated, bounded spatio-temporal object of arbitrary size. But I assert there actually isn’t a good reason to think there is some clean definition of a world boundary beyond this concept of a causally connected region or patch. (Please note I’m not saying physicists/cosmologists need to adopt my definition of a world or universe, this is just to help make sense of a philosophical problem.) If one asserts a world contains a concrete infinity, then one has exceeded the boundary for what can fit in a world. This is the error.

Now, let’s go back and consider the question of whether there is a foundational “level” of reality using this model of an infinite space of metaphysical possibilities broken down into finite worlds. When considering a world, its “parts” have a good claim for being basic given indivisibility. Since a world can have an irregular and changing boundary as events move into or out of causal contact with the center, the “whole” of the world seems to have less of a claim to be foundational.

On the other hand, let’s consider all of reality, i.e., the total space of metaphysical possibilities (all possible worlds). In this case, there exists an infinite number of ways to parse it, and the parts no longer seem to have as good a grip on being basic. Here it is the total space which is the ultimate source of the reality individual worlds take part in. So in the bigger picture, the monistic whole takes priority, and has the best overall claim to be the foundational entity.