p-zombies are inconceivable. With notes on the idea of metaphysical possibility.

by Massimo Pigliucci

Philosophy of mind and the nature of consciousness are fascinating topics, which recur both here at Scientia Salon [1] and at my former writing outlet, Rationally Speaking [2]. And of course we can hardly talk about consciousness for long before running into one of the most famous (and, in my mind, pernicious) thought experiments in philosophy of mind: the philosophical zombie! [3] (Now you should hear ominously sounding music in the background…)

In this essay I propose to do the following: we are first going to take a look at Chalmers’ zombie argument (one of a number of instances of zombification in philosophy of mind) to see exactly what it says and why I find it utterly unconvincing. Next, we’ll use p-zombies to broaden the discussion to parse the differences among different types of possibilities, especially logical, metaphysical and physical/nomological (with a nod toward two other types: epistemic and temporal/historical/contingent). Finally, we’ll use whatever we think we have learned in the process to talk even more broadly about the very nature of philosophical inquiry — or at the least the sort of analytic-type [4] metaphysics that Chalmers and his supporters indulge in.

What, if anything, are p-zombies?

The first thing to be aware of when we talk about p-zombies is that the concept has a long history and has been used in different ways for different purposes. It can be traced back to Saul Kripke’s arguments in the 1970s against type-identity theory in philosophy of mind as presented in his Naming and Necessity [5]. Versions of it were elaborated upon during the same decade by both Thomas Nagel and Robert Kirk. But we won’t go into any of that, focusing instead on the more famous Chalmers’ version of 1996.

Also to keep in mind is the fact that p-zombies have been deployed against different targets, the most important of which are behaviorism [6] and physicalism [7], though they have also been used to argue for the existence of an epistemic gap in our understanding of consciousness (the so-called “hard problem” [8], very much related to the issue of physicalism). Anti-behaviorism p-zombies are not particularly interesting, in my mind, because they beat a dead horse: while behaviorism survives as a useful practice in psychology, it has gone the way of the dodo as a theory of mind decades ago. Besides, it seems to me trivially true that one could face a being that behaves like a human and yet is internally constructed very differently from a human. So what?

As for the epistemic problem concerning consciousness, if by that one simply means that we don’t yet fully (or even very partially!) understand how consciousness can be produced in a physical system, that is obviously the case. But such observation licenses precisely nothing about the future likelihood, or even the in-principle possibility, of such knowledge. Accordingly, we will set this one aside as well.

Instead, I’ll tackle the best known deployment of p-zombies, by Chalmers, whose target was physicalism itself. Here is his argument, formalized:

1. According to physicalism, all that exists in our world, including consciousness, is physical.
2. Thus, if physicalism is true, a metaphysically possible world in which all physical facts are the same as those of the actual world must contain everything that exists in our actual world, including conscious experience.
3. We can conceive of a world physically indistinguishable from our world but in which there is no conscious experience. It follows that such a world is metaphysically possible.
4. Therefore, physicalism is false. (Because 4 follows from 2 and 3 by modus tollens.)

As plenty of commentators have noted, Chalmers’ argument is deductively valid, i.e., if its premises are true, then the conclusion must logically follow. You would think so. Any philosopher caught putting forth an invalid argument should have his license revoked (well, philosophers don’t really have licenses, but you get my drift).

Which means that we need to focus on (1), (2) and (3) above, from which (4) is allegedly derived. (1) is a fair statement of the metaphysical notion of physicalism, so no problem there. (2) is also true, pending further elucidation of what one means by the tricky term “metaphysically possible.” (3) is where Chalmers hits what will turn out to be an insurmountable brick wall.

Let’s look at it more closely, highlighting the potentially problematic bits: we can conceive of a world physically indistinguishable from our world but in which there is no conscious experience. It follows that such a world is metaphysically possible.

What does it mean to be able to “conceive” such world, in this case? And what, exactly, is metaphysical possibility?

The pertinent dictionary definition of “conceiving” is: to form a mental representation of [something]; to imagine. Well, in that sense, I can certainly conceive of a p-zombie, and of a lot of other nonsensical objects as well. Heck, I can fool myself that I can even conceive of a squared circle, which is mathematically impossible. Clearly, something more rigorous has to be in play here for the argument to go through.

One thing that is not in play is physical possibility. Chalmers himself clearly stated: “Zombies are probably not naturally possible: they probably cannot exist in our world, with its laws of nature.” [9] He also wrote: “it certainly seems that a coherent situation is described; I can discern no contradiction in the description” [10] by which I assume he means no logical contradiction. And finally: “From the conceivability of zombies, proponents of the argument infer their metaphysical possibility” [11].

Keep the above distinction among logical, physical and metaphysical possibilities in the back of your mind for a bit, it will turn out to be crucial.

There are, of course, a number of standard responses to Chalmers’ argument. Dan Dennett famously thinks that Chalmers is simply confused about the notion of conceivability: p-zombies are not actually conceivable, because when examined more closely they embed hidden contradictions, just like many other imaginary objects that we are free to conjure up in our mind when we impose few if any constraints. The better response, in this case, came from AI researcher Marvin Minsky, who pointed out that Chalmers’ argument begs the question: in order to work, it has to assume that some characteristics of human beings, namely consciousness, are not produced by physical processes, which is precisely what the argument allegedly sets out to prove. In formal terms, there is a hidden premise in the argument, lurking somewhere between (2) and (3) above. Once this premise is brought to light, the argument is actually shown to be circular (though still logically valid, mind you!).

What is metaphysical possibility?

Which brings us back to the pesky idea of metaphysical possibility. What is it, exactly, and how is it distinct from logical and physical possibility? This is critical, because Chalmers’ whole gambit relies on the proposition that metaphysical possibility is, in fact, distinct from the other two forms, and doesn’t collapse into one or the other. As we shall see, this assumption may be hard to defend.

Let us begin with a little classificatory scheme of possibilities (and, equally, impossibilities):

• Logical: typically thought of as the broader kind of possibility (and, conversely, the strictest kind of impossibility). “My friend Phil is a married bachelor” is an example of logical impossibility, as the word bachelor is defined as an unmarried man. More intriguingly, it turns out that something called Noether’s theorem [12] mathematically proves that if the laws of physics are invariant through time then the principle of conservation of energy has to hold true. That is, it is logically impossible for a universe to have both time-invariant laws and violations of conservation of energy. (Naturally, it also follows that it is not logically contradictory, as far as we can tell, to imagine a universe whose laws change over time and where the conservation principle does not hold.)
• Metaphysical: this is the pesky one. It is commonly thought to be either co-extensive with logical possibility, or to be a subset of it. Perhaps the best known example of metaphysical necessity (which is, of course, stronger than mere possibility) is Kripke’s statement that “water is H2O” is metaphysically but not logically necessary. The idea is that there is no logical contradiction is thinking that water might not be H2O (so this doesn’t rise to the level of logical necessity), and yet it seems physically impossible for that to be the case. Hold this particular thought in place while we turn to the next level.
• Physical (or nomological, i.e., law-mandated): this means that something is possible (or impossible, or necessary) under the laws of physics of our universe. For instance, for me to jump out the window and start levitating is logically possible (i.e., the idea involves no apparent contradiction of any principle of logic), and yet it is physically impossible. Philosophically, what is interesting here is that ever since Hume, many philosophers have treated the laws of physics as metaphysically contingent, in the sense that they could have been otherwise. Indeed, if the concept of the multiverse [13] holds water, this is likely the case. Some philosophers, however, like Sydney Shoemaker [14], think that the laws of physics couldn’t be otherwise, in which case physical and metaphysical possibility would turn out to be the same.

As I mentioned above, there are two other types of possibilities that occasionally enter into these discussions: epistemic and temporal/historical/contingent. Let’s examine them briefly before setting them aside and going back to the all-important concept of metaphysical possibility.

Temporal (or historical, or contingent) possibility simply means that things could have been otherwise not only logically, but also physically (i.e., without contradiction of either logical principles or physical laws). For instance, I could be writing this essay in my office at CUNY’s Graduate Center, rather than at home, as in fact I am. While contingency plays a crucial role in pretty much all the so-called “special” sciences (i.e., anything but fundamental physics), it doesn’t really concern us here. [15]

As for epistemic possibility, this is marginally relevant to our discussion, as hinted at above. For instance, at the moment at least, we do not know whether Goldbach’s conjecture [16] in mathematics is true or not. But, presumably, there is a fact of the matter as to whether it is or it isn’t, regardless of whether we will or will not ever be able to find out.

Back to metaphysics, then. First off, remember that metaphysical possibility lies somewhere between logical and physical possibility. But here is a new question: which logic, and which physics?

The question is not facetious, and indeed I suspect may lead us to a pretty solid conclusion of our quest, which will imply the rejection of Chalmers’ argument above, among other things.

As any above-introductory level student of logic knows, there are many types of logics out there [17], so that, strictly speaking it makes no sense to say that something is logically possible/impossible unless one also specifies which type of logic applies. For instance, in classical Aristotelian logic, the principle of non-contradiction applies: either A is B, or A is not B, but definitely not both [18]. But if one adopts dialethic logic [19] it turns out that certain propositions (typically, those involving logical paradoxes under classical logic) can be both true and not true. (That said, unlike my colleague Graham Priest, I don’t actually buy this, but it will do for the sake of illustration here.)

Analogously, we need to be clear on which set of laws of physics we are operating. The obvious one is the set that applies to our universe, but one could imagine situations applicable to other parts of the multiverse (if it exists) that are governed by different laws. Indeed, some physicists have even proposed that the laws of our universe change over time [20], which means that we would have to specify not just which universe we are talking about, but also when (again, personally not buying it, but it’s out there). Consider again Kripke’s assertion that “water is H2O” is not logically, but metaphysically, necessary, because it cannot physically be otherwise. Well, that’s true for our laws, in this particular period of the universe’s evolution. But if one or more physical constants were different, then it would be perfectly (physically) possible for a substance to have the chemical constitution H2O and yet for its behavior to be different from what we call “water.”

The bottom line is that I think that when people talk about metaphysical possibility/impossibility/necessity they can do so coherently, and position metaphysical space in between logical and nomological spaces, only because they are not careful enough to specify which logic and physics they are contemplating. I suggest further that if we are more precise and fix a particular logic and a particular nomology, then metaphysics space collapses into either one or the other (i.e., some metaphysical possibilities/impossibilities/necessities will turn out to be logical, and other, physical).

So, what about analytic metaphysics, then?

All of the above would seem to bode badly for the whole idea of doing anything like analytic metaphysics, the sort of thing Chalmers engages in [21]. Indeed, one could use the above analysis to question the point of philosophical inquiry itself, at least conducted in the above fashion.

Except, of course, that what you just read is an essay in philosophy — not science, not logic, not mathematics, not literary criticism. What we did was to begin with a startling claim, which was apparently presented in a logically solid framework; we parsed it, unpacked it, poked into it, until we found one or two weak points; we then proceeded to explore just how weak those points were and why; and we finally reached some conclusion, which of course is going to be tentative and open to the very same process, which I’m sure my readers will launch into with gusto as soon as this is published. That’s philosophy, folks!

What about metaphysics and metaphysical possibility, however? If it is true, as I’ve argued above, that metaphysical possibility/necessity/impossibility collapse into either their logical or physical counterparts, depending on the particular statement being made and on which logic and physics one considers while making it, what is left to do for metaphysicians?

Here I endorse (with some caution) the now famous approach to metaphysics articulated in a number of fori by James Ladyman, Don Ross, and colleagues [22]. Although these authors refer to what they do as “scientific” or “naturalized” metaphysics, and they openly disdain the Chalmers brand, labeling it as “neo-Scholasticism,” in reality, I hope, they are all doing the same thing (though some better than others). What they are doing is positioning metaphysics at the interface between logic and science, from which it naturally follows that metaphysicians will arrive at conclusions that are constrained by either the principles of logic or the laws of physics, depending on which side of that spectrum of possibilities they happen to be focusing. P-zombies, as it turns out, are physically, and therefore metaphysically, impossible (in this universe), though not logically so. But the fact that they are logically possible is no argument against physicalism at all, since physicalism itself isn’t a logical necessity, it just happens to be the best (in the specific sense of most fecund) metaphysical approach to understanding the world.

_____

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] Is quantum mechanics relevant to the philosophy of mind (and the other way around)?, by Quentin Ruyant, SciSal, 21 July 2014; The Turing test doesn’t matter, by Massimo Pigliucci, SciSal, 12 June 2014; Information is the new Aristotelianism (and Dawkins is a hylomorphist), by John Wilkins, SciSal, 1 May 2014.

[2] Three and a half thought experiments in philosophy of mind, by Massimo Pigliucci, Rationally Speaking, 6 September 2013; Computation, Church-Turing, and all that jazz, by Massimo Pigliucci, Rationally Speaking, 5 August 2013; Philosophy not in the business of producing theories: the case of the computational “theory” of mind, by Massimo Pigliucci, Rationally Speaking, 29 July 2013; The zombification of philosophy (of mind), by Massimo Pigliucci, Rationally Speaking, 29 July 2008.

[3] The Conscious Mind: In Search of a Fundamental Theory, by David J. Chalmers, Oxford University Press, 1996.

[4] Is there (still) a continental-analytic divide in philosophy?, by Massimo Pigliucci, SciSal, 17 July 2014.

[5] Naming and Necessity, by Saul A. Kripke, Wiley-Blackwell, 1972.

[6] Behaviorism, by George Graham, Stanford Encyclopedia of Philosophy.

[7] Physicalism, by Daniel Stoljar, Stanford Encyclopedia of Philosophy.

[8] What hard problem?, by Massimo Pigliucci, Philosophy Now, 2013.

[9] Consciousness and Its Place in Nature, by David J. Chalmers, in: The Blackwell Guide to the Philosophy of Mind, edited by S. Stich and F. Warfield, Blackwell, 2003, p. 5.

[10] The Conscious Mind, p. 96.

[11] Consciousness and Its Place in Nature, p. 5.

[12] Noether’s Theorem.

[13] The multiverse as a scientific concept, by Coel Hellier, SciSal, 3 June 2014 & 5 June 2014.

[14] Time Without Change, by Sidney Shoemaker, Journal of Philosophy, 1969, 66:363–381.

[15] Of course, if metaphysically you are a reductive determinist then temporal possibility reduces to physical possibility, because nothing could have happened differently, and our sense that there are stochastic events in the universe is the result of limited epistemic access to reality.

[16] Goldbach Conjecture.

[17] Logics, by John Nolt, Cengage Learning, 1996.

[18] Aristotle on Non-contradiction, by Paula Gottlieb, Stanford Encyclopedia of Philosophy.

[19] Beyond true and false, by Graham Priest, Aeon magazine, 5 May 2014.

[20] Time, laws, and the future of cosmology, by Lee Smolin, Physics Today, 2014, 67:38-43.

[21] Metametaphysics: New Essays on the Foundations of Ontology, ed. by David Chalmers, David Manley and Ryan Wasserman, Oxford University Press, 2009.

[22] Scientific Metaphysics, ed. by Don Ross, James Ladyman and Harold Kincaid, Oxford University Press, 2013.