Metaphysics and (lack of) grounding

by Massimo Pigliucci

I must admit to always having had a troubled relationship with metaphysics. My first exposure to it was during my three years of philosophy in high school (in Italy), where the bulk of our exposure to metaphysics came down to the medieval Scholastics (of course, we also studied Aristotle and Descartes, among others). The Scholastics still have a bad reputation in philosophical circles, where the very term “Scholasticism” is a polite synonym for mental masturbation, despite the fact that medieval logicians actually did excellent work (think William of Ockham and Buridan, to name just a couple) [1,2]. As a teenager prone to (intellectual) rebelliousness, though, I couldn’t but reject the Scholastics.

A bit later on, in college, I discovered (logical and empirical) positivism, which dealt yet another blow to my regard, such as it was, for metaphysics. The positivists, (in)famously, applied their verification principle to establish not just whether a given notion was true or false, but even to determine if it made sense or was rather incoherent [3]. Metaphysical concepts (such as that of God) cannot be verified, so they are literally meaningless, not even wrong. I loved it!

Jump a number of years forward, when my interest for philosophy began to rekindle while I was in the midst of my career as a scientist, and there comes David Hume and his famous fork [4]:

“If we take in our hand any volume; of divinity or school metaphysics, for instance; let us ask, Does it contain any abstract reasoning concerning quantity or number? No. Does it contain any experimental reasoning concerning matter of fact and existence? No. Commit it then to the flames: for it can contain nothing but sophistry and illusion.”

Again, such a neat and no nonsense kind of view, and from one of my favorite philosophers! Nonetheless, I tried to keep an open mind about metaphysics as a viable project within modern philosophy by, among other things, reading a book on the subject at least once in a while [5].

Which, of course, soon led to the next blow, delivered by James Ladyman and Don Ross’ difficult but highly rewarding Every Thing Must Go [6]. It proposes a different model for metaphysics, what they call “scientific” or “naturalized” metaphysics. The idea is that metaphysics can not longer be conceived as a search for a priori truths about the world, because the only reliable sources of such truths is (a posteriori) empirical evidence (and therefore science), aided by mathematics and (to a lesser extent) logic. Indeed, Ladyman and Ross tellingly label the standard way of doing metaphysics “neo-Scholasticism” (which they assuredly don’t mean as a compliment), referring to their own approach as neo-positivist in spirit (thus acknowledging both that there was something profoundly wrong with the original positivism, but also that there were valuable insights to be preserved and built upon).

Ladyman and Ross conceive metaphysics as a philosophical project directed at examining how fundamental physics and the so-called “special sciences” (i.e., anything but fundamental physics) hang together, how they can be interpreted as offering a coherent view of the world despite very disparate methods and findings (think, for instance, quantum mechanics vs economic theories).

Again, this very much appeals to the scientist in me, while at the same time maintaining, in modified form, a core component of philosophical inquiry: metaphysics gets to go on as a field, but properly re-conceived (just like, I might add, philosophy, which has constantly reinvented itself over the centuries).

I could stop here, except for the fact that I keep occasionally reading “standard” (i.e., non Ladyman &  Ross style) metaphysical papers or books, and going to standard metaphysical talks, such as one I attended a few days ago at CUNY’s Graduate Center, delivered by Rutgers University’s Jonathan Schaffer [7], and entitled “The Ground between the Gaps.”

Schaffer has been thinking (for about a decade, according to his own recalling) about “grounding,” a relatively new concept that has gotten metaphysicians (or, according to some, metaphysicists) all excited [8]. I have tried and failed to understand grounding for a while now, and specifically how it differs, if at all, from causality. Relatedly, I have tried, and again failed, to understand the idea of metaphysical truths (or necessities) as distinct from either nomological (law based, from science) or logical ones [9]. These are precisely the two topics that Schaffer’s talk set out to clarify, so I eagerly, if somewhat skeptically, seated myself in the first row and took notes. As it turns out, my skepticism was justified.

But I’m getting ahead of myself. Schaffer began by reminding his audience of the “explanatory gap” between the physical and the phenomenal: we just don’t understand (yet) how masses of cells interconnected in certain ways inside our brains generate “the feeling of what happens,” to put it as philosophically savvy neuroscientist Antonio Damasio once did [10].

Notoriously, there are two major ways to understand the gap: one is to say that phenomenal consciousness is somehow the result of brain activity, but we haven’t figured out how, as of this writing; another is to say that no matter how much we are going to learn about the brain, there is an unbridgeable divide between physicality and consciousness. I have no problem at all with the first way of looking at things, and I refer to the second one as the Chalmers delusion, after notorious “philosophical zombies” guy David Chalmers [11].

Schaffer seemed to buy into Chalmers-type explanatory gaps, and indeed claimed that they are not limited to phenomenal consciousness, but are rather all over the place. Even supposedly uncontroversial cases of reduction, such as from chemical to physical theory, are metaphysically problematic, they are not “transparent,” to use Schaffer’s terminology, and they therefore require some sort of “opaque” metaphysical bridging principle to account for them.

Now, typically the metaphysical concept invoked in these cases is that of supervenience [12], where the relations between lower and higher level phenomena is such that no change in the latter can occur without some (to be specified) change in the former. For instance, consider the money I have in my pocket at this moment: at a lower level of analysis, it consists (say) of two 25-cent coins, one 1-dollar bill, and two 20-dollar bills; at a higher level of analysis I have a total of $41.50 in my pocket, regardless of the “mechanistic” details. The higher level description obviously has a one-to-many relation to the lower level one, since I could have, say, six 25-cent coins, no 1-dollar bills, and two 20-dollar bills. This would still amount to a total of $41.50, but the makeup of the phenomenon at the lower level would be different in the two cases. The relation of supervenience says that no matter how the total is instantiated in terms of coins and bills, the only way to change the total value of $41.50 is to alter the composition of the system at the lower level. Another way to say this is that the total is (very, in this toy example) weakly emergent from the ensemble of the lower level constituents.

The problem is, according to Chalmers and others, that you can’t get phenomenal experience from neurobiology this way, some more robust metaphysical relation is needed. Which is why Chalmers invokes dualism and Schaffer calls on grounding. I think Schaffer’s approach is more sensible than Chalmers’ (unless one is a radical reductionist [13], which I am not), so let’s see where he goes with it.

Schaffer’s next move was to explain that grounding plays in metaphysics the same role that causality plays in science: just like physical phenomena are caused by other physical phenomena, metaphysical concepts or principles are grounded (i.e., explained, accounted for) by more basic metaphysical concepts or phenomena [14].

Now, how would we know whether an attempt at physicalist reduction is bound to fail (as opposed to simply having failed so far), indicating the need for grounding? Schaffer here gave the same (disappointing) answer that Chalmers is infamous for: conceivability. If it is conceivable, say, that there could be a being that is made exactly like me, atom per atom, and who however doesn’t experience any phenomenal consciousness, then this is sufficient to show a lacuna in physicalism. The difference between Chalmers and Schaffer, again, appears to be in how they seek to fill such a lacuna.

But I reject the very idea that conceivability is a reliable guide to metaphysics at all. It is too vague a concept, too flexible, and too prone to error. I’m sure someone out there can conceive of squaring the circle, for instance, and yet we know that’s a fantasy [because π turns out to be a transcendental number: 15]. My response to the p-zombie argument [9,11] is simply that they are logically possible (only because logic imposes very loose constraints on reality) and physically impossible. I don’t recognize an intermediate category of metaphysical possibilities as distinct from the other two.

Still, let’s continue with Schaffer’s entertaining and enlightening (about the status of classical metaphysics) talk. At this point he made an interesting move: following up on his analogy between grounding and causality, he proposed that we could deploy the metaphysical equivalent of structural equation modeling (SEM) [16].

SEM is a very highly developed set of statistical techniques, especially useful in biology and the social sciences, which allows us to test causal models against empirical observations. It works by constructing alternative “path diagrams” [17] specifying different possible causal relations among a given set of variables, some of which may even be unobservable directly (so called “latent” variables, like fitness in evolutionary biology, or intelligence in psychometrics). One then extracts covariances among empirically observed variables and uses them to test a set of alternative causal models, hopefully pruning a number of them from the realm of viable possibilities. SEM, incidentally, is also the technique that shows why, contra popular opinion, one can — under certain specific circumstances — infer causation from correlation [18].

What does all of this have to do with metaphysics? Schaffer suggests that any time we are faced with a problem of reduction such as that of chemical to physical theory, or of psychology to biology, we can build a SEM-type model where the causal relations among variables are replaced by different types of grounding relations (yeah, this is a bit fuzzy, but it gets very technical very quickly, though to his credit Schaffer did provide a couple of examples, mostly having to do with mereology [19]).

The (big) problem is that I don’t think the analogy is going to work, at all. You see, actual SEM is based on two components: on the one hand the causal model(s) to be tested; on the other hand the actual variance-covariance matrices summarizing the observations and telling us how the variables are statistically related to each other. Assuming that Schaffer can get a bit less fuzzy about how exactly he is going to replace causal links with grounding relations (I’m doubtful he can), he’s still faced with the fact that he has no “data” to “test” his models.

I actually asked him about this specific point in the q&a following the talk, and he made some gestures toward “counterfactual covariation,” meaning, I assume, that one could generate “data” from thinking about how two variables are related across possible worlds (using modal logic [20]). But this won’t do. Besides the fact that Schaffer would have to get other metaphysicians to agree that his particular way of generating counterfactual covariation is acceptable (good luck with that!), he would still be left, in my opinion, with nothing like SEM-style variance-covariance matrices, and therefore with no way to actually carry the analysis through, except by handwaving. Which metaphysicians can do even now, with no need to invoke either SEM or possible worlds (well, actually a good number of them do invoke the latter).

I also asked Schaffer a second question during the q&a: could he please provide me with just one convincing example (by which I mean nothing to do with theology) of metaphysical necessity? He couldn’t. He tried, of course, but all he could come up with was to outline a way, using his approach, in which such an outcome could be arrived at. Which wasn’t what I asked. If even one of the leading metaphysicians who has thought long and hard about grounding cannot answer a simple direct question like that one, I’m afraid my skepticism about the current status of standard (as opposed to “naturalized”) metaphysics still appears more than justified. Hume docet.

_____

Massimo Pigliucci is a biologist and philosopher at the City University of New York. His main interests are in the philosophy of science and pseudoscience. He is the editor-in-chief of Scientia Salon, and his latest book (co-edited with Maarten Boudry) is Philosophy of Pseudoscience: Reconsidering the Demarcation Problem (Chicago Press).

[1] William of Ockham, SEP entry.

[2] John Buridan, SEP entry.

[3] Logical Empiricism, SEP entry.

[4] An Enquiry Concerning Human Understanding, by D. Hume, Wiki entry.

[5] For instance: What is this thing called Metaphysics?, by B. Garrett.

[6] Every Thing Must Go: Metaphysics Naturalized, by J. Ladyman and D. Ross.

[7] Jonathan Schaffer’s philosophy page.

[8] Metaphysical Grounding, SEP entry.

[9] p-zombies are inconceivable. With notes on the idea of metaphysical possibility, by M. Pigliucci, Scientia Salon, 4 August 2014.

[10] The Feeling of What Happens: Body and Emotion in the Making of Consciousness, by A. Damasio, Houghton Mifflin Harcourt, 1999.

[11] What Hard Problem?, by M. Pigliucci, Philosophy Now, issue 99, 2013.

[12] Supervenience, SEP entry.

[13] Reductionism, IEP entry.

[14] A full paper by Schaffer on grounding as analogous to causality is: Grounding in the Image of Causation, final pre-publication draft; to appear in Philosophical Studies.

[15] Squaring the circle, Wiki entry.

[16] Structural equation modeling, Wiki entry.

[17] Though technically path analysis is a subset of SEM: Path analysis, Wiki entry.

[18] See the excellent Cause and Correlation in Biology: A User’s Guide to Path Analysis, Structural Equations and Causal Inference, by B. Shipley.

[19] Mereology, SEP entry.

[20] Possible Worlds, SEP entry.