Anti-Physicalist Arguments: A Tier-List

A guest post by Brian Cutter

The following is a guest post from the philosopher Brian Cutter. Brian’s an excellent philosopher—good at forming beliefs on most subjects other than the truth of Catholicism. He’s one of the inventors of both the psychophysical harmony argument and the nomological harmony argument. Read more of his papers here.

---

Physicalism about the mind is the view that all your mental properties (feeling pain, experiencing color, thinking about trigonometry, wanting chocolate, etc.) are identical to or fully constituted by physical properties. The claim is not merely that your mental states “arise from” physical activity in your brain/body. The claim is rather that all facts about your mental life consist in some set of physical facts—facts about the arrangement, movements, and interactions among bits of matter in your brain/body (perhaps together with physical stuff in your environment). For the physicalist, the relationship between the mental and the physical is not like the relationship between smoke and fire. Smoke arises from, but is nonetheless “ontologically distinct” from, the underlying fire. Instead, it’s more like the relationship between a traffic jam and the movements of individual cars, or the relationship between a stampede and the movements of individual animals. A stampede doesn’t merely “arise from” the movements of lots of individual animals. The stampede is, or is constituted by, those movements.

Physicalism is the correct view about many domains. We should all be physicalists about cartwheels, for example. A cartwheel is constituted by a complicated pattern of movements, spatial relationships, and causal interactions among physical body parts and the ground. It would be hard to characterize the pattern precisely, but even without a precise physical analysis, it’s clear that a cartwheel is nothing over and above physical occurrences. The same goes for rainstorms, avalanches, volcanoes, quasars, and earthquakes. A tad more controversially, physicalism is probably true about biology as well. Processes like digestion, respiration, reproduction, and metabolism seem to be fully constituted by the movements, spatial relationships, and mechanical interactions of material things. (I think this would be true even if it turns out that non-physical “vital forces” or entelechies push around the atoms inside living things. Digestion itself is a wholly physical process even if some of its causes are non-physical, just as a blizzard is a wholly physical event even if it’s caused by a ghost.)

While physicalism is true in many domains, I do not think it is true about the mental domain. This post gives an extensive overview of arguments against physicalism about the mind (hereafter, just “physicalism”). Kids these days like tier lists with grades A-F, often with plus/minus grades (and an S for reasons I don’t understand). My judgment isn’t so finely discriminating, so I’ll go with a coarser ranking in terms of strong arguments (a lot of evidential force), medium-grade arguments (a moderate amount of evidential force), and weak arguments (little, no, or negative evidential force).

Strong Arguments

The distinctness intuition

This is not so much an argument as an intuition, but the intuition is extremely powerful. For many, I suspect it serves as the most fundamental source of resistance to physicalism. It just seems intuitively clear that an experience—say, a phenomenal presentation of neon green—is distinct from any physical state of your brain/body. No matter how tight the correlation is between neon-green experiences and a given physical state, it just seems obvious that it’s one thing to have a neon green experience, and another thing to have so-and-so neural firing pattern in your visual cortex (or to be disposed to behave in such-and-such way, or to implement so-and-so computation, or whatever).

Here’s an analogy: it seems intuitively plausible—indeed, it seems utterly self-evident to the light of reason—that being round is not the same thing as being red. Now, you might have some empirical evidence for this distinctness claim. You might have encountered red things that aren’t round or round things that aren’t red. But empirical disassociations aren’t required to know that redness isn’t roundness. Even if we found ourselves in a world where red and round were perfectly correlated—where every red thing is round and every round thing is red—it would still be intuitively obvious that being round is one thing, and being red is another. Maybe they are lawfully connected, but they clearly aren’t the same thing. In this scenario, philosophers with a taste for desert landscapes might attempt to reductively identify redness with roundness, or vice versa. Those philosophers would be wrong, and their wrongness would be prima facie obvious. As a dialectical maneuver, the non-reductionist might respond to the reductionists by giving elaborate arguments. They might appeal to conceivability (surely you can conceive of a round thing that isn’t red), or knowability (Mary* could know the shape facts without knowing the color facts), or explanatory gaps (there is no intelligible connection between the shape facts and the color facts). But the fundamental source of their non-reductionism would have less to do with these fancy arguments, and more to do with the basic distinctness intuition.

Likewise, mutatis mutandis, for many anti-physicalist philosophers. Even so, the fancier arguments can significantly bolster the case by revealing non-obvious costs of denying the basic distinctness intuition (e.g., the need to reject otherwise well-motivated principles linking conceivability and possibility).

Epistemic arguments

These arguments start with an epistemic gap between the physical and the phenomenal (e.g., the lack of intelligible connection, the conceivability of varying the phenomenal truths while holding fixed the physical truths, etc.) and then make an inference to an ontological gap. A deeper discussion can be found in David Chalmers’ paper, “Consciousness and its place in nature,” and in his book (which I highly recommend), The Conscious Mind: In Search of a Fundamental Theory.

The explanatory gap argument

1. There is an explanatory gap between any collection of physical truths and the occurrence of a conscious experience.

2. If so, then no collection of physical truths fully constitutes/grounds the occurrence of a conscious experience.

3. If no collection of physical truths fully constitutes/grounds the occurrence of a conscious experience, then physicalism is false.

4. Therefore, physicalism is false.

Let’s say there is an explanatory gap between the A facts and a fact B iff it’s unclear why, given that the A facts obtain, B should obtain as well. Start with cases where there seems not to be an explanatory gap:

B = a traffic jam occurs

A = detailed facts about behavior/movements/interactions among individual cars.

B = there is a protest

A = facts about lots of individuals chanting, marching, holding signs, etc.

B = there is a stampede

A = facts about the movements and interactions of lots of individual animals

B = this stuff is a liquid

A = descriptions of the weak intermolecular bonds between loads of individual H2O molecules, allowing them to slide freely past one another.

B = Digestion occurs.
A = detailed facts about the mechanical and chemical breakdown of food by the coordinated activities of enzymes, acids, muscle contractions, and absorption processes within the digestive tract.

In all these cases, there is a fully intelligible connection between the A facts and B. When one fully grasps the A facts, there is no mystery as to why B holds. It is clear why and how the A facts constitute the B fact. Of course in many such cases we can’t get an adequate mental handle on all the low-level details, so we can’t always see the intelligible connection. But plausibly, there nonetheless exists an intelligible connection, one that could be seen by an idealized superintelligence, like Laplace’s demon or C.D. Broad’s “mathematical archangel.” As these examples suggest, outside the domain of consciousness, the relationship between low-level and high-level physical facts typically does not involve an explanatory gap.

Note the contrast with consciousness. There does not seem to be a similar intelligible connection between physical brain activity and the occurrence of an experience, or the fact that we have a reddish rather than a greenish experience. There’s all this complicated physical processing in the meat-computer inside my skull, but why doesn’t that processing go on “in the dark”? Why the hell are the lights on? It’s deeply unclear, indeed mysterious, how physical interactions among neurons could constitute a vibrant experience of red, in a way that it’s obviously not unclear or mysterious how the movements of lots of individual animals could constitute a stampede, or how the movements and interactions of lots of molecules could constitute a process of digestion. This is the motivation for premise (1).

Premise (2) is supported by the fact that, as the examples above illustrate, when some low-level facts constitute/ground a high-level fact, there typically is an intelligible connection. The physicalist can say that this occurs in all cases of constitutive dependence except consciousness, but without some further story this looks like special pleading.

Premise 3 is true by the definition of physicalism.

The conceivability argument: Zombies and Inverts

“‘Tis an establish’d maxim in metaphysics, That whatever the mind clearly conceives, includes the idea of possible existence, or in other words, that nothing we imagine is absolutely impossible” —Hume

Take some conscious person—say, Matt Adelstein. Let “P” abbreviate a complete physical description of Matt at the current moment. P includes all details about the sizes and shapes of his various material parts, plus their spatial relationships and movements over time, plus all relevant causal/mechanistic facts (how these parts push and pull on those parts, etc.). If you like, P can also include details of how physical stuff in his environment impinges on his body/brain. We argue as follows:

1. It’s conceivable that someone has physical profile P without being conscious, or while having experiences different from Matt’s actual experience—e.g., color inverted experiences.

Note: “conceivable” here means ideal positive conceivability. The claim is that an ideal intellect could clearly and distinctly imagine this scenario, imaginatively filling in all relevant details, without any incoherence. That is, this ideal intellect could form a coherent positive conception of this scenario, without finding that it involves any internal contradiction. It’s not like trying to imagine a married bachelor, or a legless man doing a spinning roundhouse kick, or a pattern of spatial relationships that violates triangle inequality, or one of those grim reaper scenarios that Joe Schmid writes about.

2. If this scenario is conceivable, then it’s possible.

3. If this scenario is possible, physicalism is false.

4. Therefore, physicalism is false.

Premise (3) is unproblematic. Physicalism entails that Matt’s physical properties fix all his positive properties. But if zombie Matt or inverted Matt are possible creatures, then Matt’s actual phenomenal properties (the fact that he is conscious, or has a reddish experience) aren’t fixed by his physical properties.

Premise (1): apart from its face-value plausibility, it can be supported by a few lines of thought. First, it follows from a very plausible general conceivability principle:

Structural-Phenomenal Gap: For any set of truths A that are entirely about “structure and dynamics” (roughly, truths about the spatiotemporal and causal/mechanical relationships among items, plus changes in these relationships over time), it’s conceivable that A holds without consciousness.

Since physical truths are truths about structure and dynamics, premise (1) would follow from this general principle. In many cases, we can know general conceivability principles along these lines. E.g., I think we can know that for any set G of purely geometric truths, it’s conceivable that G holds without anything exhibiting a color quality like neon green. The principle above seems just as plausible.

A second argument for (1) is that the only remotely plausible way that (1) could be wrong is if phenomenal concepts admit of functional analysis (“the state that plays such-and-such causal role vis-a-vis physical inputs, physical outputs, and other internal states”). This was David Lewis’s view. But it’s easy to see that this “analytic functionalism” is wrong. Consider a robotic functional isomorph of Matt, with a silicon computer in place of Matt’s lovely squishy carbon-based brain. Or consider a China-body functional isomorph of Matt, where the people of China send radio signals to one another in ways that functionally duplicate the signaling patterns in Matt’s brain. Regardless of whether such systems would actually be conscious, it’s obviously coherently conceivable that such functional isomorphs exist without consciousness. But analytic functionalism entails that the hypothesis that these systems aren’t conscious isn’t just false, but involves a conceptual contradiction, like the hypothesis that Matt is a round square, or does jumping jacks while lacking arms and legs. This implication is clearly false, so we can reject analytic functionalism.

Premise 2: The motivation for (2) is that conceivability is a good guide to possibility, at least in a certain restricted class of cases to which this one belongs. Kripke allegedly taught us that there are conceivable (or seemingly conceivable) impossibilities, e.g., that Hesperus isn’t Phosphorus, that water isn’t H2O, or that heat isn’t molecular motion. But it’s well known that Kripkean cases don’t cast doubt on the inference from conceivability to possibility in the present case. Consider the alleged necessity that heat = molecular motion. It might falsely seem to us as though molecular motion could have occurred in the absence of heat. This is a modal illusion, Kripke thinks, which arises because we can imagine a genuinely possible scenario in which there is molecular motion that isn’t felt as heat. But we can’t explain away the apparent possibility of brain activity without experience in the same way. The analogous strategy for consciousness would say: when we seem to imagine, say, C-fiber firing without pain, we are registering a genuinely possible scenario in which C-fiber firing occurs, but it isn’t felt as pain. But pain is essentially a feeling, so if there is a possible scenario where a brain state occurs without an associated feeling of pain, this is simply a possible scenario in which the brain state occurs without pain.

The other conceivability argument: ghosts

This one relies on the conceivability of ghosts, which are the mirror image of zombies. While your zombie counterpart has all your physical properties while lacking your experiences, your ghost counterpart has all your experiences while lacking any physical properties (no body, brain, or physical environment). We can argue as follows:

1. Your ghost counterpart is conceivable.

2. If your ghost counterpart is conceivable, then your ghost counterpart is possible.

3. If your ghost counterpart is possible, then certain popular forms of physicalism are false.

4. Therefore, certain popular forms of physicalism are false.

To see why premise (1) is plausible, note that a ghost scenario is exactly what you’re imagining when you consider Descartes’ evil demon scenario, where a demon is feeding you misleading sensory impressions of a material world, but really there are no material bodies at all. You’re imagining yourself as a ghost—where all your actual experiences exist, but they don’t correspond to any material reality. This scenario seems coherently conceivable. That’s why we can’t rule it out with absolute certainty even upon careful a priori reflection (even if we shouldn’t lend it significant credence).

Premise 2: See the motivation for premise 2 in the zombie/invert argument above.

Premise 3: The possibility of ghosts rules out “certain popular forms of physicalism,” in particular, any view that identifies experiential properties (e.g., feeling pain) with broadly physical properties that essentially require one to have a brain or a body. This would include views that identify experiential properties with neural properties like having firing C-fibers (a non-physical thing could not have firing C-fibers), as well as most role-functionalist views that identify experiential properties with functional properties, so long as the functional properties in question essentially involve causal relations to physical inputs and behavioral outputs. (Maybe certain computationalist reductions of consciousness aren’t ruled out by the argument above, since computation in a non-physical substrate is possible, but I suspect variants of the ghost argument could rule out these views as well.)

The Knowledge Argument

Mary grows up in a black and white room, never experiences color, but learns all the physical facts about human color vision through black-and-white textbooks and video lectures.

1. Mary knows all the physical facts about human color vision.

2. Mary doesn’t know all the facts about human color vision. In particular, she doesn’t know what it’s like to see red. She’s ignorant of the qualitative character exhibited by the experiences other people have when they look at tomatoes, fire trucks, etc.

3. Therefore, there are non-physical facts.

Premise 1: stipulated in the case.

Premise 2: one reason to accept this is that, when Mary leaves and sees a tomato for the first time, it seems that she learns something new. She’ll say, “woah, cool, I never knew that seeing red was like this.” She won’t say, “yup, I already knew it’d be exactly like this.” But I think the point is intuitively clear even if she never learns what it’s like. This is the situation we are in with respect to, say, bats, or pigeons who see novel colors. Even if we learn all the physical details of their nervous systems, that won’t put us in a position to know what their experiences are like qualitatively.

One response says that when Mary leaves the room, she doesn’t acquire any new information. According to David Lewis, she just gains a new ability (e.g., the ability to imagine red things or recognize red things by sight). She gains “knowledge how,” not “knowledge that.” This response looks wrong on its face, since she would naturally express her newfound knowledge with a “knows that” construction. (“Wow, cool, I never knew that seeing red was like this!”) Similarly, you could know all the physical details of a robot functionally identical to yourself while failing to know whether it has color experiences that are qualitatively the same as your own (i.e., while failing to know that it does, and failing to know that it doesn’t).

The strongest response is the old-fact/new-mode response. It says that Mary learns an “old fact” under a new mode of presentation. It’s like coming to learn that Phosphorus is a planet when you already knew that Hesperus is a planet. Plausibly, [Hesperus is a planet] is the same fact as [Phosphorus is a planet], but one can know this fact under the “Hesperus” mode of presentation without knowing it under the “Phosphorus” mode of presentation. Likewise, maybe [seeing red involves subjective quality Q] is the same fact as [seeing red involves brain process B], because subjective quality Q is the same property as brain process B. But this single fact can be cognized under two different modes of presentation or sets of concepts—a physical/scientific mode of presentation (“brain process B”) and a phenomenal/introspective mode of presentation (“subjective quality Q”).

The problem with this response is that, in general, when one knows a fact under mode M1 without knowing it under M2, this is because one is ignorant of a background fact that establishes a link between the two modes of presentation. For example, maybe the Hesperus mode of presentation is “brightest heavenly body in the evening” and the Phosphorus mode of presentation is “brightest heavenly body in the morning.” If one knows that Hesperus is a planet without being in a position to know that Phosphorus is a planet, it is because one is ignorant of a further fact linking the two modes of presentation—viz., that there is a single thing that shines brightly in the evening and the morning. If one knew all facts, one would know this linking fact, and one could know the Planet fact under the “Phosphorus” mode of presentation too. Applying this model to Mary’s case would require that she lacks knowledge of at least some fact—at least the fact that links the physical/scientific mode of presentation to the phenomenal mode of presentation. Since she knows all the physical facts, this unknown linking fact would itself have to be a non-physical fact, so we get the same result: there are non-physical facts.

Matt has a nice defense of this argument here if you’re hungry for more.

The Inconceivability Argument

This is my favorite epistemic-gap argument, which I develop in a paper called “The Inconceivability Argument.” There I argue that anyone who endorses the conceivability argument is committed to endorsing this one, though the converse doesn’t hold. Notably, the usual moves for blocking conceivability-possibility inferences (e.g., the phenomenal concept strategy) don’t vitiate this argument.

1. It is inconceivable that experiential truths are wholly grounded in physical truths. (E.g., just like it’s inconceivable that something should be yellow wholly in virtue of being round, it’s inconceivable that something should have a phenomenally yellowish experience wholly in virtue of its physical features, e.g., the sizes, shapes, movements, and spatial relations among its parts.)

2. If it is inconceivable that experiential truths are wholly grounded in physical truths, then experiential truths aren’t wholly grounded in physical truths.

3. Therefore, experiential truths aren’t wholly grounded in physical truths.

I’ll refer the reader to the paper linked above for further discussion and defense.

The Personal Identity Argument (for substance dualism)

The personal identity argument for substance dualism basically says that the only way to accommodate certain highly plausible claims about personal identity over time is to accept the existence of souls (roughly, immaterial subjects of consciousness). So we should accept souls.

Classic personal identity arguments for substance dualism can be found in the work of Thomas Reid and, more recently, Richard Swinburne. But in my view the cleanest and strongest formulation is given in Mike Huemer’s excellent introductory philosophy book, Knowledge, Reality, and Value. Here’s a simplified presentation that loosely follows Huemer:

1. The following are some true claims about personal identity:

1. Diachronic Identity: People persist through time (e.g., I am identical to someone who ate breakfast this morning).

2. One-one: Identity is a one-to-one relation, not a one-many relation. You can’t be numerically identical to two distinct people.

3. Transitivity: If A is identical to B and B is identical to C, then A is identical to C.

4. Intrinsicality: Who a given being is depends entirely on facts about that being. It does not depend directly on facts about other beings. “You cannot, for example, end a person’s existence solely by creating another person with certain characteristics who never interacts with the original person.”

5. Objectivity: “If A is a person and B is a person, there is an objective fact as to whether A = B. It is not subjective, indeterminate, or a matter of convention whether I exist in a given scenario.”

2. If the above claims are true, then the soul theory of personal identity is true, since it is the only account of personal identity that can accommodate all of them. (The soul theory says that a person X who exists at t2 is numerically identical to a person Y who exists at t1 iff X at t2 has the same soul as Y at t1.)

3. Therefore, there are souls. (This follows from the soul theory plus (a) above.)

Let’s start with premise (2): Briefly, no-self Humean views are inconsistent with (a). The most straightforward psychological-continuity and physical-continuity views are inconsistent with (b). To illustrate, suppose you walk into the teletransporter; it scans your brain/body, destroys the original brain/body, sends the data from the scan to two other machines, where two molecular duplicates of the original body are created. If we say that a later person X = an earlier person Y iff the later person is psychologically continuous with the earlier person, we get the result that you are identical to both of the resulting people, violating One-One.

There’s a patch to the theory that avoids this result. We can add a “no branching” clause. Roughly, a person X at t2 = a person Y at t1 iff (i) X at t2 is psychologically continuous with Y at t1, and (ii) no one else at t2 meets condition (i). But this violates Intrinsicality. Relatedly, it gets crazy results in non-destructive teletransporter cases. As part of a routine checkup, your eccentric doctor asks you to step into a machine for a quick scan. When you step in, your body is scanned, but not destroyed. You walk out and head to lunch. Meanwhile, unbeknownst to you, the data is sent to another machine which creates a molecular duplicate of your body on the other side of the world. According to the patched theory, in this scenario you cease to exist. Neither person is you, because both fail condition (ii). This is absurd.

As Parfit argued, condition (e) is likely to fail for any “reductionist” theory of personal identity, i.e., any theory that says that personal identity over time consists in certain more particular facts about psychological and/or physical connections between the earlier and later person. On any such theory, there will be edge cases where it’s indeterminate, or a matter of convention, or a matter for linguistic stipulation, whether the relevant psychological or physical connections are strong enough for the later person to count as “the same person.” (Parfit supports this through considering a spectrum of cases that gradually weaken psychological/physical connections. See Part 3 of Reasons and Persons.)

Premise (1): I think each of (a)-(c) wear their plausibility on their sleeve for anyone who possesses the concept of numerical identity. As for claim (d), while the general claim that identity is intrinsic is somewhat obscure, all we really need for the argument is the quoted gloss above (“You cannot [...] end a person’s existence solely by creating another person with certain characteristics who never interacts with the original person”), and this is immensely plausible. Probably (e) will be the most controversial. Parfit famously denied it. But he emphasized that denying (e) is extremely counterintuitive. When it comes to non-persons—ships, watches, clubs, nations—facts about identity over time often seem indeterminate, dependent on linguistic convention, or otherwise non-substantive. If I disassemble a watch, then reassemble it while replacing ¼ of the original pieces, there is no substantive/determinate fact as to whether it’s the same watch. It’s a matter for linguistic stipulation; different communities could legitimately adopt different conventions without error. But it seems crazy to hold this view about your own identity over time. Imagine you are about to undergo an operation that will replace some large-ish fraction of your brain with new materials. Someone will wake up and live a happy life after the operation. Will it be you? Consider this question first-personally. It seems clearly wrong to say: “There’s no substantive or determinate fact about whether it will be you. We could call him you, or not. Different communities could adopt different conventions without error. But hey, as a personal favor, if you want to go on living, I’ll rally the folks in our community to adopt the “call him you” convention. What’s that? You’d prefer that it not be you, because the post-operation guy is going to be tortured? Okay, in that case, I’ll rally for the opposite linguistic convention. You have nothing to worry about!”

While Parfit recognized the counterintuitive implications of rejecting (e), he denied it because he thought (correctly) that accepting it requires something like substance dualism, and he thought there was insufficient evidence for substance dualism. My view is that if substance dualism is required for the truth of (e), this fact is itself evidence for substance dualism. Also, I think there are independent reasons to accept substance dualism, including some arguments below, plus the following very inviting Garden Path from Property Dualism to Substance Dualism: (i) Consciousness is fundamental. (ii) It’s natural to think that fundamental properties would belong to fundamental entities. (iii) A simple soul is a better candidate for a fundamental entity than a body or brain composed of zillions of elementary particles.

Sensory Quality Arguments

Sensory quality arguments basically say that sensory phenomenal properties essentially relate you to sensory qualities like color and shape (for example, your visual phenomenology when you see a round ball essentially involves a relation to the qualities red and round), but no plausible physicalist view can accommodate this fact, so physicalism is false. I associate this style of argument most with Adam Pautz, who gives arguments of this form in various places (e.g., in this paper and his excellent book, Perception), though I’ll frame things a bit differently than he does. (By the way, I highly recommend that you read more of Pautz’s work. In my view, he’s among the top two or three best living philosophers of mind, and he’s probably done more than anyone to shape my views on these issues. Also worth noting, since this blog has promoted the psychophysical harmony argument for theism, that he’s the guy who introduced the puzzle of psychophysical harmony in its contemporary guise (see e.g. this paper).) These arguments are powerful, I think, but (fair warning) they are a little trickier to get a handle on than the arguments above.

As a warmup to an argument I actually endorse, let’s consider an argument that I’m ultimately not inclined to accept, but which has some force and sets the stage for the real argument. This is the argument from sense data.

1. Sense data theory is true.

Sense data theory says that when one has an experience as of a red round thing (whether in a normal case or hallucination/illusion), this is a matter of being acquainted with a red and round sense datum—a mental image that is literally red and round.

2. If sense data theory is true, then physicalism is false.

3. Therefore, physicalism is false.

Premise (2) is hard to deny. When you hallucinate a red round thing, there is no red round item in your environment. And there is no physical red round item in your brain. Maybe there is some neural structure or firing pattern that correlates with the red-round experience, which in some sense “encodes the information” that a red and round item is present before you. But that neural structure is not itself red and round. It doesn’t instantiate the sensory qualities that phenomenally appear to you. It would be cool if, when one experiences a (say) a neon green star, the brain generated a physical neon green star and placed it on a screen in the middle of the brain. But that’s not how it works. (Citation needed.) So, if you’re aware of some item that is actually red and round, it must be a non-physical item.

Premise (1) will be more controversial, but it has some initial plausibility. Once you reject the maximally naive direct-unmediated-acquaintance-with-the-world model of perception on the basis of the usual considerations, it’s natural to suppose that perception involves a kind of indirect awareness of outer things. We are indirectly aware of external objects in virtue of our direct awareness of the mental images they produce in our minds, images that exhibit the colors and shapes that phenomenally appear to us. Perhaps the strongest motivation for the sense data theory is that it’s the only reasonable way to accommodate what Adam Pautz calls the Pricean intuition. H.H. Price writes,

“When I see a tomato there is much that I can doubt. I can doubt whether it is a tomato that I am seeing, and not a cleverly painted piece of wax. I can doubt whether there is a material thing there at all [...] One thing however I cannot doubt: that there exists a red patch of a round and somewhat bulgy shape, standing out from a background of other colour-patches, and having a certain visual depth, and that this whole field of colour is presented to my consciousness … that something is red and round then and there I cannot doubt … that it now exists, and that I am conscious of it—by me at least who am conscious of it this cannot possibly be doubted.”

I myself don’t accept the sense data theory. I don’t confidently reject it either, but I lean toward the intentionalist alternative that would reject the Pricean intuition (despite the Cartesian certainty Price claims for it). That is, I would reject the assumption that experiences as of red round things involve acquaintance with things that are actually red and round. The intentionalist says that, in cases of hallucination/illusion, there only phenomenally appears to you to be something red and round. (My preferred version would take phenomenal appearance as a primitive propositional attitude, i.e., a primitive relation to propositional contents.)

But I think we can rehabilitate the argument by relying on a much more modest claim that will be congenial to both sense data theorists and intentionalists. Namely: the visual phenomenology associated with a red-round experience entails at least the appearance of something red and round before you. From here, it can be argued that no relevant physical property has these entailments, so your visual phenomenology can’t be identified with any physical property.

We can put the argument like this. Let R be the fully specific phenomenal property that characterizes your total visual phenomenology on some particular occasion when you experience a red round thing right in front of you (normal conditions, good lighting, etc.).

1. Having R entails having an appearance of something red and round. That is: necessarily, for any individual S, if S has visual phenomenology R, then it appears to S that there is something red and round before him.

2. None of your physical properties, except perhaps certain of your extrinsic/historical properties, entail having an appearance of something red and round (or, indeed, any significant relation to the qualities red and round).

3. Phenomenal properties aren’t extrinsic/historical properties.

4. Therefore, phenomenal properties aren’t identical to physical properties.

Premise 1 seems self-evident. Consider what it’s like for you as you look at a red round thing. Now try to imagine someone else, whose total visual phenomenology is exactly like this, but where it doesn’t even appear to him as though there is something red and round. This just seems clearly inconceivable.

For premise 2, let’s limit our attention to intrinsic physical properties, like internal brain states. One argument for 2 relies on epistemic-gap reasoning, but that would make the argument too similar to the epistemic arguments above, so let’s set this motivation aside. The interesting thing about this argument is that it can run without any controversial inferences from epistemic to ontological gaps. The key point is just that, among your intrinsic neural properties, none seem to entail, none essentially involve, any interesting relations to the sensory qualities red and round. Granted, at least for roundness, you do have certain physicalistically acceptable relations to external instances of roundness. You are in a neural state that is normally caused by round things, a state that perhaps has the function of being produced in response to round things (a function rooted in your evolutionary history, perhaps). These would be the standard physicalist explanations of how it is that you manage to be in a state that represents, or “encodes the information” that, something round is before you. But these are all extrinsic and historical properties. If we set those aside, and we just look at physical properties you’d share with (say) swampman, or a Boltzmann brain, none of those physical properties entail any interesting connection to the qualities round and red. For any relevant intrinsic neural property, we can fully spell out what it is to have that property without mentioning terms like red and round—say, simply in terms of arrangements and firing rates of neurons.

Premise (3) is meant to rule out a standard physicalist approach to accommodating premise (1). This is the externalist intentionalist view of Michael Tye, Fred Dretske, William Lycan, and others. Roughly, the thought is that to have a phenomenal property like R is to be in a neural state that represents that there is a red round thing, which is then reductively cashed out in terms of having a neural state that “tracks” redness and roundness. Tracking is an extrinsic historical relation to worldly properties. Roughly, a neural structure tracks roundness iff the visual system has the function of producing that neural structure in response to round things, or something along these lines. Functions are then cashed out in terms of selectional history. The result is that your phenomenology constitutively depends on extrinsic and historical facts.

I find this view impossible to believe. It’s reasonable enough to say that the “information encoded” or “represented” by a neural structure depends on historical and extrinsic facts. But the idea that my current phenomenology—what it’s like to be me right now—constitutively depends on extrinsic historical facts just seems crazy.

This flatfooted intuitive case for (3) is sufficient for me. But if you want to be hammered over the head with tons of empirical evidence for (3) from neuroscience and psychophysics, I’ll refer you to some papers by Adam Pautz. (This, this, and this.)

Medium-Grade Arguments

Material composition arguments

These arguments start with some claims about mereology (the study of parts and wholes) and derive anti-physicalist conclusions with the help of some auxiliary assumptions.

The argument from mereological nihilism

In a recent interview I did with Joe Schmid, Joe said that Dustin Crummett accepts an argument along the following lines, which I think has some force:

1. Mereological nihilism is true. That is, the only things that exist are simple things, things without proper parts. There aren’t any composite objects.

According to the mereological nihilist, strictly and metaphysically speaking there aren’t any chairs. There are only particles arranged chairwise. Similarly, there aren’t human bodies or brains, just particles arranged bodywise/brainwise. It’s fine to talk as though there are such things, but this is just a useful fiction, just as we might talk as though there are sakes (“he did it for John’s sake”) or absences (“he was alarmed by the absence of food in the fridge”) while thinking that, strictly speaking, such things don’t exist.

2. I exist.

3. I’m not a simple physical thing.

4. Therefore, I’m a simple non-physical thing—a soul.

Premise (3) should be unproblematic. I’m not an elementary physical particle like an electron.

Premise (2) is something I can know with Cartesian certainty, just by walking myself through Descartes’ cogito argument.

Premise (1) (mereological nihilism) will be the most controversial premise. But there’s a surprisingly good case for it. First, it gives a simple, elegant answer to the “special composition question,” namely: if you’ve got two or more things, what would you have to do to them to get them to compose a further thing? The nihilist says: nothing you could do to them would make them compose a further thing because composition never occurs. The other simple/elegant answer is universalism: every plurality of things composes a further thing, no matter how scattered or seemingly unrelated they are (so, e.g., there is a thing composed of your pinky toe, my nose, and the rings of Saturn). If you don’t give one of these two answers, you face some awkward questions. Do the original things have to be touching (and how do we define touching)? Do they need to all behave in a unified way (and what exactly does that mean)? Or do they need to causally interact with each other (and in what ways)? Going down this road leads to all sorts of trouble, e.g., complicated, arbitrary-looking, anthropocentric composition principles and vagueness about what exists. Best to go with either nihilism or universalism.

Does nihilism have advantages over universalism? Well, it’s simpler. Relatedly, it avoids causally redundant objects. In order to explain why the glass shatters, we only need the particles arranged baseball-wise. We don’t need to posit a further thing—a baseball—that those particles compose. So Ockham’s razor tells us not to posit it. You might think universalism does better at respecting common sense, since it gives us chairs, cars, mountains, and so on. But that’s not so clear. First, it gives us lots of things that common sense disavows (like the object composed of my nose plus your steering wheel). Second, it doesn’t automatically give us ordinary objects without further hefty assumptions. Universalism tells us that there is a car-shaped thing composed of the particles in my driveway. Is it a car? For all universalism says, maybe it’s just a car-shaped “aggregate of particles,” something that is essentially composed of exactly those particles, something that will become a weirdly scattered object once I replace a few parts of my car. If so, that’s not a car as ordinarily conceived, since cars can gain and lose parts over time. So maybe there are two perfectly co-located things, one with the persistence conditions of an “aggregate of particles,” one with the persistence conditions of a car. This is already pretty weird, since you might have thought two things can’t occupy the exact same location at the same time. But it gets weirder, since there’s not going to be a principled reason to stop at just two car-shaped things in my driveway. To avoid arbitrariness, we’ll probably have to say that there are infinitely many objects that coincide with my car, one for each possible set of persistence conditions. Arguably we’re better off accepting nihilism.

The argument from nihilism or universalism

Personally, I’m not fully sold on the argument above for nihilism. I’m also open to universalism (with infinitely many co-located objects everywhere). But there’s another argument that starts with the disjunction:

1. Either (mereological) nihilism is true or (mereological) universalism is true.

From here, we can argue that nihilism + physicalism entails too few subjects of consciousness, while universalism + physicalism entails too many subjects of consciousness, so we should reject physicalism.

2. If nihilism and physicalism are true, then either I don’t exist or I’m an elementary physical particle.

3. I exist, and I’m not an elementary physical particle.

4. If universalism and physicalism are true, then there are multiple conscious subjects in my vicinity with more-or-less the same experiences as me.

5. There aren’t multiple conscious subjects in my vicinity with more-or-less the same experiences as me.

6. Therefore, physicalism is false.

The motivation for (1) and (3) are given above. (2) is true by the definitions of nihilism and physicalism. (5) is highly plausible, part of common sense, and denying it would likely lead to massively revisionary moral views (any action you think is causing one person pain is actually causing many people pain).

That leaves us with (4). The basic argument for (4) goes something like this: if physicalism is true, I’m some physical thing, maybe a brain or a body. Whatever physical thing I am, if universalism is true, there will be countless other things that mostly coincide with me, but differ in whether they include this or that peripheral atom as a part. These things will be almost physically indistinguishable from me. So if consciousness is a physical property, they should be conscious just like me. If we say otherwise—if we say that they aren’t conscious, but we insist that consciousness is nonetheless a physical property—then we’d have the absurd result that a miniscule physical difference of a single peripheral atom doesn’t just cause, but constitutes, the difference between having rich, vivid, technicolor phenomenology and having no experience at all. (For a much more careful version of this argument and some other arguments like it, see my paper, “The Many-Subjects Argument against Physicalism.” There I argue that we also get too many subjects when physicalism is combined with a “conservative” ontology, which only accepts only “ordinary” common-sense objects. For related arguments see Peter Unger’s “The Mental Problems of the Many” and Jon Simon’s “The Hard Problem of the Many.”)

The argument from free will

1. We have free will.

2. Free will is incompatible with physicalism.

3. Therefore, physicalism is false.

We’ll say that you have free will iff (on some occasions) the following two conditions hold:

• Multiple possibilities: there are multiple courses of action available to you, i.e., you have the ability/power to perform each of two or more actions (taking “action” broadly to include omissions/refrainings).

• Control: you have control over which of these actions you perform.

Premise 1: Arguably premise (1) has the status of a Moorean truth (“one of those things we know better than we know the premises of any philosophical argument to the contrary”). The idea that there are sometimes multiple courses of action open to us, where it’s up to us which of these we pursue, is utterly central to our commonsense view of persons. Maybe (1) is also supported by introspection. Some say that, when one introspects on the process of decision making, it appears to one that there are multiple courses of action available to oneself, and that one has control over which of these is performed. Insofar as we have presumptive reason to trust appearances (on pain of radical skepticism), this would give us presumptive reason to accept (1). Finally, we can support (1) through arguments like this:

1. Some people (e.g., Nazis) have done things for which they are blameworthy.

2. If S is blameworthy for doing X, it was within S’s power to refrain from doing X, and S had control over whether s/he did X.

3. Therefore, some people have free will (as defined above).

Arguments like this raise lots of tricky issues that I won’t get into here, but the premises look prima facie plausible. Certainly the free-will denier will have to say some pretty awkward things about either the conditions on moral responsibility or the blameworthiness of (e.g.) Nazis.

Premise (2): many people (including most of my undergraduate students) find it intuitively plausible that physicalism rules out free will. The basic intuition might be put like this: if physicalism is true, I’m just a complicated collection of atoms, each of which is just following a script provided by the laws of physics, laws that I did not create and over which I have no control. Since my choices and behavior, on physicalism, are ultimately constituted by the behavior of my constituent atoms, it seems to follow I have no ultimate control over my choices or behavior.

We can also motivate (2) with the ol’ determined-or-random dilemma. On the one hand, maybe the ultimate physical laws are deterministic. Then given the distant past (over which I have no control) and the laws (over which I have control), there’s only one physically possible way for the future to unfold. That certainly sounds like it’s in tension with the idea that there are multiple courses of action that are genuinely open to me. On the other hand, maybe the laws are indeterministic, with some subatomic randomness thrown in. And maybe some of those random subatomic events are brain events that partly constitute my choices. But since I don’t have control over random subatomic events, this wouldn’t give me real free will either (maybe it creates multiple possibilities, but the control condition doesn’t hold). (Some will say that the determined-or-random dilemma rules out free will for everyone, even for non-physicalists. I disagree. If I have a primitive power to choose A, and a primitive power to choose B, and I primitively exercise the first power, though I could have exercised the second, this suffices for multiple-possibilities and control. This is importantly different from the physicalist scenario where my choice is a macroscopic physical occurrence constituted by a bunch of random subatomic events. If my choice is “determined from below” by a bunch of individual indeterministic events, that threatens freedom just as much as if my choices are “determined from behind” by past events.)

Here is a related argument, which is similar to one given by Peter Unger (in this paper), for the incompatibility of free will and physicalism. (See also this excellent paper by Jason Turner.)

The mere-aggregate argument for the incompatibility of free will and physicalism

1. If physicalism is true, you are a mere aggregate of physical parts (cells, molecules, ultimately subatomic particles).

Say that X is a mere aggregate iff X is a composite entity (something composed of multiple parts) such that all of X’s powers are reducible to the powers of its basic parts. A car, for example, is a mere aggregate. It has various powers, e.g., the power to drive. Driving is nothing more than the parts exercising their various more basic powers in certain ways. In this way, the car’s power to drive is reducible to the powers of its parts.

2. If you are a mere aggregate, you don’t ultimately have control over your decisions—when you form an intention to do this or that, this is “determined from below” by the operation of your various microscopic parts working in tandem.

3. Therefore, if physicalism is true, you don’t ultimately have control over your decisions.

Premise 1 will hold for at least standard versions of physicalism. Premise 2 is motivated by the idea that if X is determined by some things that, individually, you don’t have control over, then neither do you have control over X.

This argument leaves at least two non-physicalist options for securing free will. First, there is the Cartesian option, where you are a non-physical being. Your powers as a non-physical being go beyond the powers of the basic material parts of your body, which is distinct from you. Second, there is a “property dualist” option: you are wholly composed of material parts, but some of your powers, including your powers of free choice, are “radically emergent,” in the sense that they aren’t reducible to the powers of your material parts.

The Determinacy Argument

1. Determinacy: It is always determinate whether a given individual is conscious.

2. Indeterminacy: If physicalism is true, it is sometimes indeterminate whether an individual is conscious.

3. Therefore, reductionism is false.

Premise (1) seems intuitively very plausible. Colin McGinn writes, “the concept of consciousness does not permit us to conceive of genuinely borderline cases of sentience, cases in which it is inherently indeterminate whether a creature is conscious: either a creature definitely is conscious or it is definitely not.” Maybe some things, like snails, have only a tiny degree of consciousness, but as long as there’s even any consciousness at all, they are determinately conscious. I agree with McGinn. I can’t even conceive of a case where it’s indeterminate whether a thing is conscious—where there’s no fact of the matter as to whether there’s anything it’s like to be that thing.

Premise (2) is very hard to deny. If you look at the physical processes underlying the transition from a state of non-consciousness to a state of consciousness (whether in the course of evolution, or in fetal development, or just in the transition from dreamless sleep to wakefulness), there is no sharp physical joint that could mark a sharp cutoff. The underlying physical features are continuous parameters—firing rates, number of neurons, complexity of nervous system, amount of information integrated, how many specialized subsystems have access to a given bit of information, amount of noise in the informational channel between this part and that part of the brain, and so on. There is simply no sharp, significant, physical transition here at all. There are countless precise physical properties in the vicinity, with different arbitrary sharp cutoff points. If consciousness is a non-physical superadded property, the laws could link one of these to the onset of consciousness. But if there isn’t an extra primitive property here—if there’s just the physical stuff—there’s no way that our use of “conscious” would determinately single out one of these precise physical properties, any more than our use of “heap” or “tall” determinately singles out one precise physical property.

(Maybe you think it’s fine to allow for at least mild indeterminacy, e.g., indeterminacy about whether a snail has a faint flicker of experience or no experience at all. But surely there couldn’t be radical phenomenal indeterminacy, e.g., a case where it’s indeterminate whether something has rich, vivid, technicolor experience just like yours vs. no experience at all. Or indeterminacy between extreme pain and zombiehood. Or indeterminacy between reddish and greenish phenomenology. It turns out that many versions of physicalism will lead to the possibility of radical indeterminacy, but that’s a more complicated argument that I won’t get into here. See Adam Pautz’s paper, “The Significance Argument for the Irreducibility of Consciousness,” for discussion of closely related points.)

The argument from rational insight

The argument from rational insight says, roughly:

1. We know lots of stuff by rational intuition, e.g., about morality (torturing others for fun is wrong), math (5+7=12), metaphysics (red and green exclude one another, parthood is anti-symmetric), logic (~(P&~P)), and epistemology (you shouldn’t have a credence in P of .9 while also having the same credence in ~P).

2. The only way we could know lots of stuff by rational intuition is if we have some non-mechanistic, non-physical, rational mode of access to abstract facts (say, because this is the only way to explain our reliability, or the only way to avoid genealogical debunking by securing an explanatory connection between the facts and our beliefs).

3. Therefore, we have some non-mechanistic, non-physical mode of rational access to abstract facts.

I like Joe Carlsmith’s description of the resulting view in the case of moral knowledge (he doesn’t endorse the view, but seems to find it at least a bit tempting):

“[T]he non-natural Good regains some small amount of the theistic God’s power. It gets to touch Nature, from the outside, at least a little, via some special conduit closely related to Reason, Intelligence, Mind. When we do moral philosophy, the story goes, we are trying to get touched in this way; we are trying to hear the messages vibrating along some un-seen line-of-contact to the land-beyond, outside of Nature’s Cave. And sometimes, somehow, the Sun speaks.”

If we identify the Sun that Speaks with God, we get something like a traditional Augustinian “divine illuminationist” view of rational intuition (a highly underrated view imo). Matt has an excellent post defending an argument along these lines, so I won’t discuss it further here. See also this paper by Tomas Bogardus, this paper by John Bengson, and this paper by me for related ideas.

The Argument from Intentionality

The argument from intentionality basically says that there is no adequate physicalist account of intentionality—the fact that our mental states are about things in the world, and have truth-conditions or satisfaction conditions—and therefore physicalism is false. I think this is right, but to defend this argument we’d need to survey a range of proposed physical reductions of intentionality and argue for their inadequacy. I’m not going to do that here, but see the first few pages of David Chalmers’ “Inferentialism, Australian style” for a great high-level overview of the issues, the main reductive approaches, and the main difficulties they face.

The Axiological Argument

The axiological argument says that experiences differ from their physical correlates in evaluative respects, and therefore must be distinct from their physical correlates by Leibniz’s law.

1. Some experiences, like excruciating pain and euphoric pleasure, have “final” value or disvalue (i.e., they are good or bad, not just instrumentally, but in themselves).

2. The physical correlates of these experiences (e.g., the associated types of neural activity, behavioral dispositions, or functional profiles) do not have final value or disvalue (at least not to the same degree).

3. Therefore, some experiences are distinct from their physical correlates.

Premise (1) is very plausible and will be accepted by just about anyone who isn’t an error theorist about goodness/badness.

Premise (2) is also plausible on reflection. The physical processes in the brain associated with euphoric pleasure or excruciating pain, considered in terms of their physical nature, seem not to be particularly good or bad in themselves. If they bring about pleasure or pain, they might be instrumentally good/bad, but nothing about the physical processes in themselves seems to be significantly good or bad. To bring out the intuitive support for this claim, consider a zombie duplicate of someone in excruciating pain. Intuitively, it seems that there is nothing very bad going on in this scenario. I claim that even if we imagined the situation in arbitrary physical detail, it would still seem that nothing very bad is going on. But note that when we imagine the zombie in full physical detail, we are considering all the actual physical correlates of excruciating pain. The intuition that nothing bad is going on in the zombie scenario reflects the more fundamental intuition that those physical correlates, considered in themselves, are not disvaluable. This doesn’t presuppose that the zombie scenario is metaphysically possible. We are only concerned with whether the physical correlates of pain are bad in themselves, and the (conceptually possible) zombie scenario includes all those physical correlates. Our judgment that nothing bad is going on in the zombie scenario reflects a judgment about the (lack of) disvalue of those physical correlates.

Another way of making the same point: for any physical profile we encounter—say, the physical profile of a robot, a detailed brain simulation, or an alien that behaves in happy-seeming ways—the following judgment seems plausible: if this physical stuff is all that’s going on, and the relevant physical processes aren’t accompanied by a subjective experience of happiness, then there is nothing very good going on with this being. That judgment reflects a more fundamental judgment that those physical processes, considered in themselves, are not valuable to any significant degree.

Weak Arguments

Descartes’ Argument from Divisibility

In the sixth Meditation, Descartes gives roughly the following argument for dualism:

1. Every material thing is divisible.

2. The mind is indivisible.

3. Therefore, the mind is not a material thing.

Premise (1) is questionable because there might be indivisible physical particles. But we could patch the argument by adding “except elementary particles” to (1) and adding an unobjectionable premise saying that the mind is not an elementary particle.

The real problem with the argument is that (2) is highly non-obvious. Descartes considers the objection that the mind has parts like the faculty of willing, perceiving, understanding, etc., and responds reasonably that distinct faculties needn’t be construed as distinct parts. Fine, but without further argument it’s hard to see how we can rule out that there are parts of the mind hidden to introspection.

An Argument from Certainty Allegedly Given by Descartes

The following argument is often attributed to Descartes’ in undergraduate philosophy classes. (I suspect it often plays the role of allowing for an easy dunk on Cartesian dualism to make the physicalist alternative look more attractive.) I myself don’t think Descartes ever gave quite this argument, but there are some passages in the second Meditation that sort of suggest it.

1. I can be certain that the mind exists.

2. I cannot be certain that the body exists.

3. Therefore, the mind is not identical to the

body.

The premises look pretty good, I think. The problem is that the argument is invalid, because it substitutes into an opaque context—the content clause of a propositional attitude report. It’s invalid in the same way as the following argument: (i) Jones believes that the evening star is a planet. (ii) The evening star is the morning star. (iii) Therefore, Jones believes that the morning star is a planet.

The Aristotelian/Thomistic Argument for the Immateriality of the Intellect

Aristotle gives a somewhat obscure argument for the immateriality of the intellect in De Anima III. Aquinas gives a related argument in the Summa:

“For it is clear that by means of the intellect man can have knowledge of all corporeal things. Now whatever knows certain things cannot have any of them in its own nature; because that which is in it naturally would impede the knowledge of anything else. Thus we observe that a sick man's tongue being vitiated by a feverish and bitter humor, is insensible to anything sweet, and everything seems bitter to it. Therefore, if the intellectual principle contained the nature of a body it would be unable to know all bodies. Now every body has its own determinate nature. Therefore it is impossible for the intellectual principle to be a body. It is likewise impossible for it to understand by means of a bodily organ; since the determinate nature of that organ would impede knowledge of all bodies.”

The two key premises here seem to be:

1. By means of the intellect, we can have knowledge of all corporeal things (in particular, things with widely varying natures---sticks, stones, horses, trees, quasars, trampolines).

2. If the intellect essentially made use of a bodily organ (as vision uses the eye), we wouldn’t have this ability; because that bodily organ would have its own nature, which would exclude it from taking on the natures of other things.

3. Therefore, the intellect doesn’t essentially make use of a bodily organ.

The argument leaves something to be desired in the way of clarity. I’m certainly no expert on Aristotle or Aquinas, but the key background idea, inherited from Aristotle’s theory of knowledge, seems to be this: when we intellectually know a thing, the nature of that thing is received in the intellect. If the intellect involved a bodily organ, that organ would have its own definite nature, which would impede it from receiving contrary forms or natures. But clearly we are not so limited in what we can know. We have an intellectual ability to grasp an open-ended, wide range of different natures. This shows that intellectual activity does not occur by the operations of a bodily organ.

As a standalone argument, I classify this as “weak,” just because it requires us to accept an elaborate, contentious, and highly non-obvious background theory about how sensing and cognizing work. Maybe this theory (or something close enough for the purposes of the argument) can be adequately defended, but in my view the background framework doesn’t look especially promising. E.g., the idea that the idea that vision involves your physical sense organs receiving sensible forms from objects looks to be in prima facie tension with contemporary vision science. While the neo-Aristotelian project of reconciling a broadly Aristotelian account of the mind with contemporary science may be worth pursuing, it’s not the tree I want to be barking up.