How to evaluate a philosophical theory
A handy-dandy guide to make sense of what philosophers say
I don’t know if you have come across the writings of my colleague Peter Adamson. I’ve recently made them the focus of a special edition of my occasional recommended readings. Peter is a professor of ancient and medieval philosophy at King’s College London and the host of a long running podcast entitled A History of Philosophy Without Any Gaps. He was also the 2021 keynote speaker at the annual World Philosophy Day event that I host at City College.
And yet, this post is not really about Peter! However, it is inspired by something he wrote at the beginning of chapter 39 of the book he co-authored with Jonardon Ganeri on classical Indian philosophy. Here is what they say:
“When you’re evaluating a philosophical theory, there are a few tests you should always carry out. Does the theory preserve our intuitions? If not, does it give us sufficient reason to prefer this new theory to our previously held beliefs? Does the theory fall prey to counterexamples or leave unexplained exceptions? And here is one more test that should never be omitted: can the theory survive being applied to itself?”
It’s an excellent guide to understanding philosophical claims, so I figured we could apply it to a couple of instances to begin to get the hang of it.
First, though, a caveat. While Adamson and Ganeri talk about “philosophical theories” I don’t particularly go for that term because it reminds people too closely of scientific theories, and I think there are major differences between the two, because science and philosophy—though continuous and mutually informing fields—are sufficiently different from each other that we shouldn’t mix up their respective methodologies.
Science works on the basis of theories that relate directly to empirical evidence. Scientific theories—say quantum mechanics, or the theory of evolution—arise in response to new evidence concerning natural phenomena, attempt to account for such evidence, and make predictions about new, as yet unobserved phenomena.
Philosophical “accounts”—my preferred term—do not depend directly on empirical evidence, though they ought not to be in contradiction with such evidence. And their aim is not to produce predictions about future observations, but rather to further a broad conceptual understanding of certain problems. Examples include accounts of ethical systems, such as virtue ethics, Kantian deontology, or utilitarianism; as well as accounts in philosophy of science, like Thomas Kuhn’s idea of paradigm shifts, or Karl Popper’s notion of the role of falsification in demarcating science from pseudoscience.
That said, we can then agree with Adamson and Ganeri about the following schema for evaluating philosophical accounts:
(1) Does the account preserve our intuitions?
(1b) If not, does it give us sufficient reason to prefer this new account to our previously held beliefs?
(2) Does the account fall prey to counterexamples or leave unexplained exceptions?
(3) Can the account survive being applied to itself?
Let’s consider two famous examples and how the Adamson-Ganeri test, as I shall call it, applies to them.
Example 1: Logical positivism
Logical positivism, which after WWII became know as logical empiricism, was an early 20th century movement in philosophy of science that attempted to re-establish philosophy itself on more scientific grounds—an enterprise that had been tried and had failed a number times before, from Plato (who used Pythagorean geometry as an ideal) to Kant (who considered Newtonian mechanics as a model).
The crucial notion underlying logical positivism was the so-called verification principle, which stated that there are only two types of meaningful statements: (i) those that can be empirically verified; and (ii) logical tautologies (which include mathematical theorems).
The verification principle was a modern restatement of a famous view articulated by the Scottish Enlightenment philosopher David Hume, and sometimes referred to as Hume’s Fork:
“All the objects of human reason or enquiry may naturally be divided into two kinds, to wit, relations of ideas, and matters of fact. Of the first kind are the sciences of Geometry, Algebra, and Arithmetic ... [which are] discoverable by the mere operation of thought. ... Matters of fact, which are the second object of human reason, are not ascertained in the same manner; nor is our evidence of their truth, however great, of a like nature with the foregoing.” (Enquiry Concerning Human Understanding, section IV, part I)
The chief target of both Hume and the logical positivists was metaphysics, and especially theology. Essentially, if one adopts either Hume’s Fork or the verification principle, theology and metaphysics go out the window, since they are concerned neither with “matters of fact” (i.e., empirical evidence) nor with “relations of ideas” (i.e., math and logic). I must admit that I’ve always been sympathetic to this notion, and to this day am very critical of so-called analytic metaphysics.
Let us see how logical positivism performs against the Adamson-Ganeri test:
Does the account preserve our intuitions? Well, that depends on whose intuitions we are talking about. I think it is reasonable to focus on the intuitions of what philosophers call “the relevant epistemic community.” If we were talking about quantum mechanics, say, that community would be constituted of physicists working on quantum mechanics, and perhaps philosophers of science interested in the same topic. The intuitions of a biologist, or of a football player, wouldn’t count—just like my intuitions about football can safely be ignored. So in the case we are discussing the relevant epistemic community would comprise philosophers of science and epistemologists. And yes, I do think our intuition that empirical evidence or logical arguments are the major, if not the only, sources of knowledge we have, is preserved.
Does the account fall prey to counterexamples or leave unexplained exceptions? No, at least not in my opinion. Of course a theologian would claim that our knowledge of god is a counterexample, but that would be begging the question, and besides, I don’t think we have any knowledge of god, only made up stories. Another objection might be based on daily examples of “knowledge,” as in “I know that my partner loves me!” Well, maybe you do or maybe you don’t. But if you, that’s because you have observed loving behavior toward you from your partner. And that’s empirical evidence!
Can the account survive being applied to itself? Here is where logical positivism, famously, fails. The verification principle cannot be applied to itself, because it isn’t a logical-mathematical statement, and there is no empirical evidence that it is true. This objection, unfortunately, goes also for Hume’s Fork: the very book from which the above quote comes, the Enquiry Concerning Human Understanding, is full of statements that are neither matters of fact nor relations of ideas, and yet it’s an excellent and valuable book on philosophy and epistemology.
The next question then becomes: can the failure of logical positivism to meet the Adamson-Ganeri criteria for a good philosophical account be fixed? That is, can we modify the original position in a way that, in this case, it is applicable to itself? I don’t think so, though people have tried (for instance, Ladyman and Ross in their thought provoking Every Thing Must Go: Metaphysics Naturalized). The best I can do is to suggest that there are good reasons to be extremely skeptical of any statement about the nature of the world (notice the qualification) that are not backed up by empirical evidence. Notice that this statement, which I could pompously christen “Pigliucci’s anti-bullshit criterion,” does survive the self test, because it is not a statement about the nature of the world, only about what we can say about the nature of the world.
Example 2: p-zombies
Philosophical zombies, or p-zombies, are a thought experiment first introduced by Robert Kirk at the University of Nottingham and then popularized by New York University’s David Chalmers. The idea is that we can imagine the existence of a being that is outwardly exactly like us and which behaves like us, but that is nonetheless incapable of conscious experience. A p-zombie, for instance, would react to being poked in the eye with all the standard outwardly manifestations of what appears to be pain, but would not actually feel pain. Even more stunningly, the p-zombie is anatomically and biologically identical to us—for instance he has a functional human-like brain—and yet does not experience any of the mental phenomena that human beings experience.
Chalmers’ argument—and this is the philosophical account we are concerned with—is that the conceivability of p-zombies demonstrates that physicalism is false. Physicalism is the notion that everything that exists in the world and has powers of cause-effect is physical, that is made of matter/energy. If physicalism is wrong, Chalmers continues, then our only option is to accept some kind of mind-body dualism in order to account for what he calls the “hard problem” of consciousness, that is the existence of first person experience.
I have argued in the past that the so-called hard problem is actually based on a category mistake, i.e., it is the result of a particular kind of conceptual confusion on the part of Chalmers and others. Plenty of critics have also advanced the objection that conceivability, which is Chalmers’ criterion for accepting the premise of his thought experiment, is a vacuous and therefore unreliable epistemic criterion. Yet other critics, like Daniel Dennett, have suggested that p-zombies are logically incoherent, but I don’t think that’s right. The notion of p-zombies doesn’t seem to me to violate any law of logic, though it does violate the established principles of physics and biology as we understand them. That is, philosophical zombies are physically (not logically) impossible in our world.
Chalmers counters that his preoccupation is not with our world but with all possible worlds, the realm of interest to metaphysics. But of course the only way to know about possible worlds is to empirically investigate the respective laws of physics, which we cannot do. The whole thing, therefore, amounts to—in my opinion—a big useless discussion, just like much of analytical metaphysics more generally.
(Notice that even if Chalmers were right and there are possible worlds in which p-zombies exist, that would somehow be the result of the laws of physics that govern those worlds. So physicalism would still be true in such worlds. And it is certainly true in our world, the only one that really matters to us.)
But we are here concerned with how p-zombies and their alleged defeat of materialism stand up under the scrutiny of the Adamson-Ganeri test, so let’s get down to it.
Does the account preserve our intuitions? No, clearly not. Again, we are talking about the intuitions of the relevant epistemic community, which in this case would be the community of neuroscientists, cognitive scientists, and philosophers of mind. The very strong negative reaction that p-zombies often elicit from members of those groups suggests that their intuitions are somehow offended. I know mine certainly are!
If not, does it give us sufficient reason to prefer this new account to our previously held beliefs? Ah, that’s the crux of the question, ain’t it? Clearly Chalmers & co. think the answer is in the positive, but again they face a large number of critics. I have outlined above a number of reasons to think that the new account is not, in fact, preferable, but you should, of course, make up your own mind about it.
Does the account fall prey to counterexamples or leave unexplained exceptions? Not exactly, though one could argue that in this case not leaving anything unexplained is actually a weakness of the p-zombie account. Since the thought experiment appeals not to the laws of physics or logic, but rather to the extremely vague notion of “plausibility,” pretty much anything goes. What is implausible to you may be perfectly plausible to me, so that the account ends up—in the minds of some—explaining too much.
Can the account survive being applied to itself? In this particular case this question is uninformative, because the p-zombie account doesn’t concern the nature of knowledge—as was the case with logical empiricism—but only a specific claim about the nature of the world (i.e., physicalism). I’m not sure what it would mean to apply the notion of p-zombies to itself.
What have we learned?
I could, of course, multiply the above examples endlessly, but this essay is already long enough. I think the Adamson-Ganeri test is very useful, both for professionals and non-professionals alike, as a framework to evaluate what I prefer to call philosophical accounts (as opposed to the too sciency-sounding “theories”). It asks good basic questions that can be used to orient ourselves around a given philosophical idea and begin to evaluate its potential merits.
That said, note from the preceding discussion that while both logical empiricism and p-zombies fail the Adamson-Ganeri test, I have also stated my opinion that the first one is far more interesting and potentially valuable than the second one. That’s because I can easily imagine modifications of the verification criterion, like the one I suggested above, that would do useful philosophical work, while I can’t imagine the use of indulging in metaphysical speculations a la Chalmers.
Of course, I could be wrong. But which criteria would you use to establish that?
I regard both the verification principle and Occam's razor [a] as good rules of thumb to detect BS, not grand, fundamental principles.
[a] Including the key caveat 'sufficient'.
You could be wrong?! Never.
Seriously, a very interesting (though slightly over the head of a non-philosopher like me) essay.