Most arguments against physicalism draw metaphysical conclusions from epistemological premises. This leaves them hostage to doubts about the epistemology. Philosophers debate the epistemology, and the metaphysics—the point, after all—too often drops from sight. This paper puts forward an argument from metaphysical premises to metaphysical conclusion. And it supports its premises, as much as possible, with metaphysical considerations. It can’t avoid epistemological issues, of course. For that would mean talking past the physicalist. But, most places, the paper recasts these issues in metaphysical terms. It asks, for example, not how we know consciousness, but how the world must work if we know consciousness the way the physicalist says we do. And then it asks how plausible that implicit metaphysics is.
1. The Argument
The paper’s argument runs as follows:
(1) The physical, as physicalism must understand it, consists entirely of extrinsic properties (relations without intrinsic relata).
(2) Conscious experience has at least some intrinsic properties.
(3) These intrinsic properties don’t reduce to extrinsic properties.
Therefore:
(4) Conscious experience has properties that can’t be physical, as physicalism must understand it.
Therefore:
(5) Physicalism is false.
Others make arguments, similar on the surface, which remain epistemological nonetheless. David Chalmers infers that physical accounts won’t explain consciousness from premises stating that (1) physical accounts explain at most structure and function; and (2) explaining structure and function won’t explain consciousness.Footnote 1 Because his premises deal with explanation rather than properties, he reaches a conclusion about the bounds of physical explanation rather than the bounds of physical nature. Gregg Rosenberg notes that science doesn’t reveal categorical properties underlying physical dispositions.Footnote 2 He reasons that the physical world therefore consists either of dispositions all the way down or of dispositions resting on empirically inaccessible grounds. Physicalism can recognize these grounds, he allows, but only as empty placeholders without intrinsic content. Rosenberg then argues that neither dispositions without non-dispositional relata nor relations anchored in empty placeholders can yield the intrinsic content of conscious experience. Unfortunately he casts this last argument in terms of non-entailment. William Seager comes closest.Footnote 3 He marshals all the elements of the argument. But he deploys them, not to attack physicalism, but to explain the explanatory gap.Footnote 4
2. The First Premise: the Nature of the Physical
2.1 Physicalism
We need a grasp of the physical as physicalism must understand it. We therefore need to know, not only what physicalism claims, but what it must claim. Physicalism holds, very roughly, that all concrete entities are either physical or nothing over and above the physical. This takes us surprisingly far when we ask how and why physicalism makes this claim.
Physicalism does metaphysics by proxy. It delegates exploring the concrete world to the science departments down the hall. It passes the torch on concrete metaphysics because it passes the torch on epistemology. After comparing the successes of science to the failures of other approaches (including, notably, a priori philosophy), physicalism judges the scientific method the best way—maybe the only way—of learning about the concrete world. Overall, physicalism accepts that science has been at it long enough, and gotten good enough, that we can—and should—trust its findings, in outline at least. It follows that physicalism loses its moorings if it distances itself from the methods and findings of science. Unless its concrete metaphysic adopts the scientific image more or less as science gives it to us, physicalism becomes underspecified and indeterminate, telling us little about the world. It has specific metaphysical content, after all, only when it takes science to have gotten it largely right. And insofar as physicalism makes metaphysical claims over and above these borrowings, or relies on non-scientific methods in its own independent investigations of concrete reality, it undercuts its warrant. For that warrant rests significantly on the success of science.
Physicalism thus necessarily inherits much from science. One bequest is metaphysical reductivism. Physicalism takes on board scientists’ working hypothesis, or empirical inference, that all non-basic entities are nothing over and above basic physical entities.Footnote 5 This gives it a bottom up rather than top down metaphysic. Physicalism thus accepts that all non-basic entities are (at least token) identical to composites or aggregates of basic physical entities tokened/instanced at the bottom level.Footnote 6 It follows, for the physicalist, that all non-basic entities derive their natures from the natures of the basic physical entities that make them up. A second bequest is non-basic mentality. Physicalism draws the empirical inference that mentality is a non-basic feature of the world. It bases this on physics’ silence about mentality (imprecise and disputed talk about a role for consciousness in the collapse of the quantum wave function aside). Science encounters unambiguous signs of mentality only when special sciences like biology and psychology investigate organisms with complex nervous systems. The physicalist therefore insists that consciousness results only when enough basic physical entities come together in the right ways.Footnote 7 This non-basic mentality claim has become the hallmark of physicalism as a mind-body theory. Its credo denies (1) that any mental properties have basic status; and (2) that mentality of any kind exists at the bottom level. These claims set physicalism off from views like emergentism (which understands conscious properties as basic but high-level properties) and panpsychism (which understands conscious properties, at their simplest, as basic and bottom-level properties).
Physicalism must also claim that high-level entities are (at least token) identical to arrangements of basic physical entities as physics describes them. Otherwise, as remarked above, physicalism undercuts its warrant and paints too indeterminate a picture of the world. It follows that physicalism must understand the physical in terms of physical theory, since nowhere else do we find the physical as physics describes it.Footnote 8 We avoid making physicalism trivially false (a risk, since current theory surely contains gaps and errors) by allowing physical theory to slide into ideal or completed theory. And we avoid emptying physicalism of content and making it trivially true (a risk, since ideal or completed theory could say anything) by (1) insisting on the non-mentality requirement; and (2) stipulating that any ideal or completed physics remain recognizably scientific—that it use the scientific method to collect data empirically and to explain them causally.Footnote 9
Doctrines that do the following therefore count, for our purposes, as physicalism:
(1) they take their ontology of basic concrete entities from basic physics;
(2) they build higher-level concrete entities out of those basic physical entities;
(3) they impose a non-mentality requirement on basic entities and on the bottom level; and
(4) they allow for extensions of basic physics that (i) deal with basic concrete entities; (ii) arise through applications of the scientific method to data collection and causal explanation; and (iii) satisfy the non-mentality requirement.
This paper thus addresses what we might call ‘standard physicalism,’ the physicalism of the overwhelming majority of physicalists. Some physicalists endorse instead a non-standard physicalism variously called ‘Kantian physicalism,’Footnote 10 ‘funny physicalism,’Footnote 11 ‘Type-F monism,’Footnote 12 ‘Russellian monism,’Footnote 13 and ‘Russellian panpsychism.’Footnote 14 Kantian physicalists accept that empirical science only reveals dispositional properties (see ahead), insist that dispositional properties have categorical grounds, and conclude that basic physical entities have an empirically inaccessible categorical nature. Most Kantian physicalists then claim that this underlying nature consists of conscious or proto-conscious properties. For five reasons this paper doesn’t consider Kantian physicalism a kind of physicalism. First, Kantian physicalism violates the non-mentality condition, the centrepiece of standard physicalist philosophy of mind. Second, Kantian physicalism recognizes entities lacking empirical warrant, which not only sets it apart from standard physicalism but also denies it the scientific warrant all physicalisms claim. Third, Kantian physicalism proposes a very different basic concrete ontology than does standard physicalism. Whereas standard physicalism (at least token) reduces all conscious properties to basic physical entities, Kantian physicalism (at least token) reduces all physical entities to basic conscious (or proto-conscious) properties. Thus Chalmers classes Kantian physicalism as a non-materialist philosophyFootnote 15 and elsewhere regards it as a kind of idealist monism.Footnote 16 Fourth, treating Kantian physicalism and standard physicalism as of a kind garbles the contrast between physicalism and panpsychism. Although some Kantian physicalists and some panpsychists share their metaphysic (hence the labels ‘Kantian physicalism’ and ‘Russellian panpsychism’), most physicalists rightly view panpsychism as an alternative mind-body theory. Fifth, because Kantian physicalism and standard physicalism have such different metaphysics, they have very different weaknesses. Arguments against them must therefore address very different issues. This paper targets standard physicalism only.Footnote 17
2.2 The Physical
What does physical theory say about the world? We answer this by noting what science does, can do and must do.
Science collects data through empirical means and then systematizes, interprets and explains them. Data thus come to the scientist by way of the senses.Footnote 18 They do so directly, through the world’s causal effects on eyes, ears, etc. And they do so indirectly, through the world’s causal effects on scientific instruments, which in turn affect eyes, ears, etc. It follows that the scientist can gather data only about dispositional properties—tendencies to affect the senses thus-and-so under such-and-such conditions.Footnote 19 The scientist then either builds her model of the world directly from her data, as does a biologist who reports the discovery of a new parrot in Borneo. Or she explains her data, by postulating properties (or entities with properties) that would cause them. In the latter case, the scientist hypothesizes new dispositions, or, at most, objects construed solely in terms of new and old dispositions. She can do no other. Hypotheses framed in terms of non-dispositional properties would lack causal explanatory power and hold aloof from empirical test. They might have metaphysical warrant, but they would lack empirical warrant. The scientist, as scientist, could not posit them. The physicalist, because she takes her concrete ontology from empirical science, could not recognize them.Footnote 20
Physics, like any science, therefore portrays the entities in its domain, which include the concrete world’s basic building blocks, as systems or structures of dispositions—dispositions arranged in space and time and related one to another. Thus physics understands the electron as a more or less durable cluster or bundle of basic dispositions to do thus-and-so under such-and-such circumstances. Physics, and the rest of the sciences, with their reductive outlook, then treat non-basic entities as composites or aggregates of basic physical entities. This makes non-basic entities complex systems/structures of interacting dispositions. It follows that all concrete entities, at least as science reveals them, consist solely of structures of interacting spatio-temporally located dispositions.
Thus science and physics say nothing about any qualities which fit into or fill out physical structure. They say nothing about any qualities which underlie, hold together or ground physical dispositions. They say nothing about the intrinsic nature, if any, of physical substance—what physical substance is like in itself, apart from its powers to interact with or relate to other things.Footnote 21 As empirical disciplines they can’t. Thus D.M. Armstrong’s plaintive words:
But if we look at the properties of physical objects that physicists are prepared to allow them, such as mass, electric charge, or momentum, these show a distressing tendency to dissolve into relations that one object has to another. What, then, are the things that have these relations to each other? Must they not have a non-relational nature if they are to sustain relations? But what is this nature? Physics does not tell us.Footnote 22
Let us use ‘intrinsic properties’ to refer to any properties a concrete entity has besides (1) its spatio-temporal (structural) properties; (2) its dispositional (causal/dynamical/functional) properties; and (3) any relations it bears to other entities not parts of itself.Footnote 23 Intrinsic properties cover the sorts of things about which science holds silent—qualities that fit into or fill out structures; qualities that underlie, hold together or ground dispositions; most of all, qualities an entity has in itself apart from its powers to interact with or relate to other things. Let us use ‘extrinsic properties’ to refer to all properties of a concrete entity other than its intrinsic properties.
3. The Second Premise: Intrinsic Conscious Nature
3.1 Experience
Conscious experience seems to have something intrinsic about it. It presents as having a determinate qualitative nature apart from its relations to, or its powers to interact with, other things. It thus appears to have a nature over and above any causal/functional properties it might also have. Braddon-Mitchell and Jackson put it this way:
Intuitively, phenomenal nature is intrinsic. The perceived redness of a sunset is a feature of how the experience is here and now; you cannot capture its nature by talking of its similarities and differences, and of what causes it, and of what it causes. It is as intrinsic as squareness. But then perceived redness in particular, and colour experience in general, cannot be captured in functional terms, in terms of causal relations.Footnote 24
Any experience has three aspects conceptually separable although perhaps metaphysically one: a conscious presentation (what is presented); a relation of conscious or subjective awareness (the presenting, which marks off experience from, say, the non-conscious ‘perceptions’ of an electric eye); and a conscious subject (what is consciously aware and thus presented to). (By ‘subject’ we needn’t mean anything grand—just what experiences, maybe just the experiencing of the experience (think of the experienced ideas making up Hume’s mind).) For two reasons we might treat these three aspects as one. First, each aspect necessitates the others. Not only can’t we have an experience with one, much less two, missing. We can’t even conceive of an experience without all three together.Footnote 25 Second, when we introspect carefully, we don’t discern two distinct entities, subject and experience, standing in some quasi-perceptual relation. We notice nothing more than a stream of experiences, each of which ‘gives’ or ‘has’ all three aspects merely through being an experience.Footnote 26 If these aspects really are the same, then the relationship between experience and subject is not that of percept to perceiver, or content to container, but of part to whole.Footnote 27 And this makes sense phenomenologically. Our conscious selves, as we experience them, consist of all and only the experiences we have.
Experience extends widely. Philosophers may attend to qualia—experienced sensory or emotional qualities (‘raw feels’). But consciousness embraces everything one has conscious awareness of. “It includes everything a bat or a new born baby can feel, and everything a great mathematician can experience in thinking.”Footnote 28
3.2 Physicalism and Experience
Physicalists divide into a priori physicalists—those who claim actual or possible a priori links between our physical and phenomenal concepts—and a posteriori physicalists—those who deny such links.Footnote 29
3.2.1 A Priori Physicalism
We saw that physicalism necessarily understands physical nature in purely causal/functional terms. It follows that physicalism necessarily construes physical concepts as causal/functional concepts. And physicalists do construe them thus.Footnote 30
Now, if physical concepts are causal/functional concepts, and if, as a priori physicalists claim, a priori conceptual links hold between physical and phenomenal concepts, then our phenomenal concepts must also be causal/functional concepts. This follows from the fact that a priori conceptual links can’t hold between purely and essentially causal/functional concepts and purely and essentially non-causal/functional concepts. But if our physical concepts and phenomenal concepts are both causal/functional concepts, then the explanatory gap between consciousness and physical nature should be a conceptual impossibility (and, assuming minimal rationality on our part, a psychological impossibility as well). This explanatory gap, recall, rests on the fact that we can’t see, and can’t see that we ever could see, how basic physical entities could result in consciousness. Now, if we already have the right determinate physical and phenomenal concepts, then, the truth of a priori physicalism granted, we should see today how physical entities result in consciousness. But we don’t. And even if we don’t have the right determinate physical and phenomenal concepts today, then, the truth of a priori physicalism granted, we should at least see the possibility of seeing how physical entities might result in consciousness. After all, no principled conceptual barrier can lie between the determinate of a currently unspecified causal/functional concept and the determinate of another unspecified causal/functional concept. So, if we have little or nothing to go on today but the fact that our physical and phenomenal concepts both qualify as causal/functional concepts, then it should seem at least possible that the entities referred to by the one might result in the entities referred to by the other. But it doesn’t. In sum, a priori physicalism entails the conceptual and (assuming minimal rationality on our part) psychological impossibility of the explanatory gap. But the explanatory gap obtains nonetheless.
The a priori physicalist might retort that conscious properties must be causal/functional entities—and hence that our concepts of them, if adequate, must be causal/functional concepts—because consciousness must have evolved and only causal/functional entities can evolve. But the non-physicalist can concede that conscious properties evolved without granting them causal/functional natures. He can claim that they evolved as the realizers of sundry functions, which, like any realizers, have properties (including, perhaps, non-causal/functional properties) under-determined by the roles they realize.Footnote 31 More, he can hold that conscious properties, although realizers of functional roles, nevertheless have purely non-causal/functional natures. Here he might claim that conscious properties ‘sit there,’ altogether passively, and physical nature, of itself, actively responds to them. This puts all active causality on the physical side.Footnote 32
Most a priori physicalists will dismiss this reasoning as too quick. They will argue that our phenomenal concepts, appearances to the contrary, either are, or should be, causal/functional concepts. The paper deals with such claims in section 3.4.
3.2.2 A Posteriori Physicalism
A posteriori physicalists deny a priori conceptual links between our physical and phenomenal concepts. They claim that we can only know a posteriori the metaphysical relation—identity, realization, whatever—that holds between physical entities and conscious properties.
Now, any concept with an actual referent must be transparent, translucent, mildly opaque or radically opaque.Footnote 33 A transparent concept reveals all of its referent’s essential properties. Thus your concept of a circle gives you a circle’s full essence: that it consists of all points equidistant from some stationary point. A translucent concept reveals some, but not all, of its referent’s essential properties. Thus your concept of a human heart reveals its functional essence (that it pumps blood) but not its physical essence (that, for example, it consists of such-and-such cells arranged thus-and-so). A mildly opaque concept reveals none of its referent’s essential properties, but at least some of its accidental properties (typically enough to fix actual world reference). Thus your concept of water identifies it as the stuff filling the rivers, lakes and oceans, etc., but doesn’t reveal its essential physical nature as H2O. And a radically opaque concept reveals none of its referent’s properties. Thus your concept <that stuff>, taken independently of whatever it refers to, functions as a blind demonstrative without ascriptive conceptual content. These four possibilities exhaust logical space.Footnote 34
Physicalists necessarily claim that our physical concepts are transparent or translucent. Recall that physicalism draws most of its warrant from the success of science, from the fact that science paints a picture of the world at least more or less right. The physicalist must therefore insist that our physical concepts reveal at least some (and likely most if not all) of their referents’ essential properties.
What about our phenomenal concepts? The a posteriori physicalist can’t allow that these are transparent. For suppose experience E has essential properties A, B, C and D. Then our phenomenal concept of E, CPHEN(E), would represent E as having A, B, C and D. More, since it truly represents the essence of E, it represents E as having A as A truly is (A qua A, rather than, say, A qua Josie’s favourite quality), as having B as B truly is, and so on. The a posteriori physicalist therefore can’t say that we have two conceptually independent grasps of each of A, B, C and D. CPHEN(E) represents A, B, C and D as, respectively, A, B, C and D. Otherwise it fails to reveal E’s essential properties, contrary to hypothesis. But the a posteriori physicalist also insists that each of A, B, C and D is a physical property. It follows that CPHEN(E) represents E as having the very physical properties it does have. But then it should be a priori that E is physical, contrary to a posteriori physicalism.
The a posteriori physicalist can’t allow that our phenomenal concepts are translucent. For suppose, again, that experience E has essential properties A, B, C and D. Suppose, also, that CPHEN(E), being translucent, represents E as having A and B. But, as above, CPHEN(E) represents E as having A as A truly is and as having B as B truly is. And A and B are physical properties. It follows that CPHEN(E) represents E as an entity having only physical properties (albeit not all the physical properties it in fact has). But then, as above, it should be a priori that E is physical, contrary to a posteriori physicalism.
The a posteriori physicalist can’t allow that our phenomenal concepts are mildly opaque. Suppose that experience E has essential properties A, B and C and accidental properties d and e. And suppose that our phenomenal concept of E, CPHEN(E), represents E as having d and e. Since d and e are only accidental properties of E, E could lack them and still be the kind of experience it is. But it isn’t plausible that our phenomenal concepts say nothing about the kinds of things they refer to. For instance, it seems that the experience of pain can’t lack all the properties our concept attributes to it, including felt hurtfulness, and still count as pain.Footnote 35 It seems that our phenomenal concepts must reveal more than mildly opaque concepts can.
The a posteriori physicalist can’t allow that our phenomenal concepts are radically opaque. In this case our concept of experience E, CPHEN(E), doesn’t represent E as having any particular properties. But then CPHEN(E) tells us nothing whatsoever about E—in particular, nothing whatsoever about what it is like to have E. But, again, it seems implausible that our concept of the experience of pain, which attributes felt hurtfulness, tells us nothing whatsoever about the experience it refers to.
It seems, in sum, that our phenomenal concepts must be either transparent, translucent, mildly opaque or radically opaque, since these four exhaust the options. But none works for a posteriori physicalism. This bodes ill for a posteriori physicalism.
Note that the above arguments attend to properties had and properties referred to. They thereby focus, as much as their target allows, on the metaphysics presupposed by a posteriori physicalist epistemology. The arguments proceed, not by directly challenging a posteriori physicalist epistemology, but rather by drawing untoward consequences out of its implicit metaphysics. The a posteriori physicalist, for his part, can resist these arguments in two ways. He can make the epistemological claim that, appearances and intuitions to the contrary, our phenomenal concepts so badly misrepresent their referents as to reveal nothing, or next to nothing, about them. In other words, he can maintain that our phenomenal concepts are mildly or radically opaque notwithstanding the arguments above. Section 3.4 addresses arguments to this effect. Or he can make the metaphysical claim that the world possesses modal structure that denies our phenomenal concepts any solid grip upon it. The next section looks at this.
3.3 Brute Metaphysical Identities
Some a posteriori physicalists postulate brute identities or brute (or strongFootnote 36) metaphysical necessities. These come down to the same thing, seen as identities when viewed metaphysically and seen as necessities when viewed modally. If A and B stand in a relationship of brute identity, then A equals B, but nothing—nothing metaphysically real—grounds an epistemically intelligible connection between them. Thus the non-identity of A and B won’t seem impossible no matter how much we know about A and B. By the same token, if A and B instance a brute metaphysical necessity, then their non-identity remains logically possible, and therefore a priori conceivable, but is nonetheless not metaphysically possible.
Brute identities/necessities have many shortcomings.Footnote 37 Here we focus on their metaphysical incoherence. We therefore view them as identities, which allows us to think of properties and the relations between properties, rather than as necessities, which, as modal entities, steer our thinking toward conceivability, rationality, and other similarly epistemological or quasi-epistemological phenomena of uncertain ontological status. So we ask how property A (a conscious property) and property B (a physical property) can stand in a relationship of identity where nothing about A and nothing about B make this intelligible—and yet where, by hypothesis, this unintelligibility results from the metaphysics and not merely from epistemology. If A and B really are the same thing—if we have just one thing rather than two distinct things somehow related—then we have a single thing with a single set of properties. Now, if the apparent bruteness doesn’t result from epistemological factors, then we don’t cognize the single property A/B as property A through one subset of its properties and as property B through another and (wholly or partly) disjoint subset of its properties. Nor do we cognize A/B as A through a subset of its properties plus some set S1 of properties it doesn’t in fact have, and as B through another (possibly identical) subset of its properties plus a different set S2 of properties it doesn’t in fact have. We instead cognize A/B as A and as B through cognizing the very same properties. More, we cognize A and B in the very same way, since the bruteness, again by hypothesis, doesn’t spring from epistemological factors. But then how can the identity or necessity be brute—and therefore unintelligible in principle? How can it fail to be straightforwardly intelligible?
In sum, brute necessities are metaphysical posits that perform two roles in a posteriori physicalist thinking. They ensure the tightest possible metaphysical link between conscious and physical properties: identity. And they account for our inability to grasp this link. But when we ask about the metaphysics presupposed by the epistemological role the a posteriori physicalist would have brute identities play, we discover that no single posit could fulfil both of the demands the physicalist places upon it.
3.4 Knowledge of Experience
Given the above arguments, the physicalist, a priori or a posteriori, must say that experience is not what it seems. Somehow our concepts and/or representations get it wrong.
3.4.1 Experience’s Hidden Nature?
Some philosophers claim that experience has hidden depths inaccessible to experience. We would know experience as physical if only we could access these depths. Thus Armstrong:
I would…advance the idea…that everything in the world, everything, every event, every property of things and events, every relation that things and events have to each other, are each one of them an epistemological iceberg. Our knowledge and rational belief about all these things, though real, is selective and limited. If you take this view then it becomes much easier to accept that the secondary qualities might have hidden depths to which we cannot penetrate in perception.Footnote 38
This makes sense only if experience includes more than we experience. But it doesn’t, of course. We can’t define a thing to include something other than itself. We certainly can’t define a thing to include something it isn’t. The thing we wish to understand is experience itself—the property of experience as experienced. So we define experience that way. If that property has hidden depths, if its roots reach into the unexperienced, then we acknowledge that experience relates to non-experiential entities, perhaps even by way of a results-from/results-in relation. But if, at the outset, we define experience to include the unexperienced, then we change the subject, from experience as experienced to an amalgam of experience and something else. We thereby risk explaining something other than what we care about: we risk explaining the unexperienced something else instead of experience itself.
Derk Pereboom offers a variant of this approach.Footnote 39 He argues that experience is straightforwardly physical. However, its mode of presentation, the way it presents to us, makes experience appear to have properties that in fact it lacks. And here Pereboom has in mind the very ‘phenomenal properties’ that, when we compare them to physical properties, give rise to the explanatory gap. But this move fares no better than Armstrong’s. If Pereboom has it right, we don’t experience experience, but instead this other thing, its mode of presentation. But then Pereboom’s ‘mode of presentation’ is nothing other than experience as we understand it, and his ‘experience’ is something else, something other than experience, something we don’t in fact experience. Pereboom must now say either that this mode of presentation has the phenomenal properties in question, in which case his re-labelling hasn’t advanced his cause. Or he must deny that any entity has them. But in that case he eliminates experience altogether, along with the phenomenal properties that make it so difficult to understand in physical terms. But this leaves it baffling why the explanatory gap was a problem in the first place—why Pereboom went to such trouble to explain it away.
In sum, philosophers like Armstrong and Pereboom argue for a relation between experience and something else. Then they explain the something else. In Armstrong’s case, the physicalist understands the something else as experience’s hidden and unexperienced essence. In Pereboom’s case, the physicalist re-labels the experience as the something else and the something else as the experience—switching their labels, as it were—and then attends to and physically explains the now unexperienced ‘experience’. In both cases the physicalist gets the epistemology wrong because he makes a basic metaphysical mistake.
3.4.2 The Immediacy of Experience
Most physicalists deny unmediated access to experience. They claim we have no better access to experience than to external objects represented in experience. They must say this. Otherwise they couldn’t easily maintain that our phenomenal concepts/representations get it wrong. And then they would find themselves pressed to take the explanatory gap at its face.
The denial of immediacy fails, however. The physicalist must say that experience involves a mediated experience of an experience—where E1, the experience the subject is conscious of, gives mediated access to E2, some underlying but nonetheless experiential state. For, if the state E2, access to which is mediated, isn’t an experience, then the physicalist doesn’t rebut immediate access to experience. And if the state doing the mediation, E1, isn’t an experience, then the physicalist denies the subject experience, eliminating experience instead of reducing it to the physical. But how do E1 and E2 relate? Note that E1, an experience, can give mediated access to E2, another experience, in such a way that the subject’s overall experience is a mediated experience, only if (1) the subject experiences as an experience E1’s representation of E2; and (2) this experiential representation of E2 is what the subject experiences. This has the result that E1 is a reproduction, facsimile or likeness of E2 (E1 can’t do what the physicalist asks and merely refer to E2 or otherwise represent E2 symbolically). It follows that E1 faithfully represents E2 only if it shares E2’s properties. But now we see that E1 can’t be the same state as E2. For E1, by hypothesis, misrepresents E2 (the physicalist, after all, denies immediacy only to claim that experience is not what it appears), and no state can be identical to an inexact reproduction of itself. (If E1 is an inexact reproduction of E2, then it represents E2 as other than E2 really is—as having properties E2 doesn’t have or as lacking properties E2 does have. But then E1, the inexact reproduction, can’t share all its properties with E2, the state it is an inexact reproduction of. It must be distinct from E2.) But if E1 is distinct from E2, and E1 is what the subject experiences, then it doesn’t help the physicalist that E1 gives mediated access to E2. If all experience is mediated, then the denier of immediacy needs somehow to mediate the subject’s access to E1. But he can do this only by introducing an E3, between the subject and E1, that does mediate E1. And this E3 must itself count as an experience (for otherwise the subject experiences nothing, again contrary to hypothesis). But now the physicalist needs another experience, E4, to mediate E3, and so on. The vicious regress ends only if the physicalist concedes some experience had immediately.
Here we’ve looked at the metaphysics implicit in the physicalist’s epistemological denial of immediacy. And we’ve found that this metaphysics involves either the existence of immediacy or the non-existence of experience. But neither of these coheres with non-eliminative physicalism.Footnote 40
3.4.3 Self-presentation and Self-revelation
It follows from immediacy that experience can’t support a distinction between reality and appearance. The experience, as it really is, and as it appears to the subject, are necessarily one and the same. This makes good sense if, as suggested in 3.1, the subject relates to his experience as whole to part rather than as perceiver to percept. In this light, we speak loosely when we say the subject has direct access to his experiences. We speak loosely even when we say he encounters them through basic acquaintance. In both cases we risk the obscure claim that the subject exists separately from his experiences while nevertheless having immediate epistemic access to them. Treating the subject as the sum of his experiences dispels the mystery. For now we see the immediacy of experience as a straightforward consequence of the metaphysics of identity.
The non-physicalist can nonetheless both acknowledge and explain our misjudgements about our conscious states. Even though experience is unmediated, our judgements, beliefs and knowledge about it are mediated. These issue from cognitive processes that take experience as input and generate, as output, judgements, beliefs and knowledge about experience. And these cognitive processes, like all cognitive processes, can err. This opens up the possibility, at the higher-order cognitive level, of a gap between appearance and reality—of a discrepancy between the first-order experience and the way the higher-order state represents it. We can sum this up by saying that experience always, and necessarily, self-presents (at the level of first-order experience), but doesn’t necessarily self-reveal (at the higher-order level involving representations of experience).Footnote 41
This paper’s arguments rely, of course, on judgements and knowledge claims about experience, and not just on experience itself. But the fact that experience self-presents places tight constraints on how badly it can fail to self-reveal. If the full experience, as it is, forms part of the conscious mind, then that mind is well-placed to know and judge it. Not surprisingly, then, most of our misjudgements about experiences happen at the margins. They typically concern very short experiences, subtle emotional discriminations, unflattering self-attributions, etc.Footnote 42 But the paper’s argument doesn’t turn on anything rare or fine. It requires only that experience have at least one intrinsic property. Yet every experience presents as having intrinsic properties, even very short experiences, subtle emotional states difficult to label, and unflattering self-attributions. Even conscious doubts about the intrinsic nature of experience themselves present with it. It follows that the physicalist, to undercut the argument, must cast doubt on the representational fidelity of experience per se.
But he can’t easily do this. Because the physicalist borrows most of his concrete metaphysic from science, he must accept empirical science as a valid way of knowing the world. He must also accept the picture painted by empirical science as a more or less complete and accurate representation of the world. But almost all scientific data reach (the cognitive domains of) our minds through consciousness. More, our explanatory theories tell us about the world only when they figure in consciousness. If we misjudge our experiences as badly as the physicalist needs, then, it would seem, we can trust none of the representations presented therein. But then we must distrust all empirical data and all theories built upon them. This means rejecting both the scientific method as a valid way of exploring nature and the scientific understanding of the world as our best model of it. But then the warrant for physicalism crumbles. The physicalist must therefore concede that self-presentation entails enough self-revelation to ground empirical science. But we need much less self-revelation than this to judge that experience involves intrinsic qualities. Whereas physicalism needs conscious representations reliable enough to accurately convey scientific information about external entities, this paper’s argument only needs minimal knowledge of experience as presented. And scientific information, theoretical and empirical, is so thoroughly enmeshed in experience that the physicalist can’t plausibly claim that experience perfectly reveals its informational content while wholly obscuring the experiential form that content takes.Footnote 43
You might object that this shows, at most, that the physicalist can’t say that the first person perspective massively misrepresents, but not that it doesn’t in fact do so. Bear in mind, however, where the burden of proof lies. Anyone who would know and understand the world must start with appearances. More, she must take appearances at face value barring specific, positive and persuasive reasons to the contrary. If the would-be theorist turns her back on appearances as such, then her project of learning and understanding could never get started. And, once started, it could never settle on a fixed picture or model of the world. Now, we have every reason to take the first person perspective at face value but for considerations put forward by the physicalist. The burden therefore lies on her to discredit its representational fidelity.
But we have good metaphysical reasons for thinking this burden very hard to meet. Earlier, the paper voiced its rejection of direct realism.Footnote 44 Before that it ventured that conscious subjects consist of their experiences. A few pages ago it argued for the immediacy of experience. These points together call to mind the following picture (sketched here in roughest outline and intended more to suggest than to ‘prove’). Conscious subjects evolved to use their experiences to represent the external world. They therefore have direct access only to an internally generated model of the world. This model serves their needs because, in adaptively relevant ways, it has features isomorphic to the world (and thus supplies information about the world). It stands to reason, then, that subjects naturally and unreflectively take (at least many of) their experiences as the external world. This nicely explains the so-called transparency of experience that philosophers like Dretske, Harman and Tye make so much of.Footnote 45 But in this case the properties of experience don’t collapse into the properties of the external world represented in experience. Instead, experience has, through identity, the very properties with which it represents the external world, and thus, in many cases at least, the very properties it represents the external world as having. From this point of view we should expect experience to do a better job revealing its own nature than the nature of the external world. It therefore shouldn’t surprise us that experience often represents the world as having qualities—such as the so-called secondary qualities—that we have trouble understanding the external world as actually having (except in the form of simulacra like dispositions to cause such-and-such experiences in us or the often disjunctive causal bases of those dispositions).
We can strengthen the point further. While experience serves mostly to represent the external world, it also monitors itself.Footnote 46 Presumably introspection evolved to help it do this. Thus we have experiential states, like pain, that may signify external conditions (such as tissue damage), and thereby alert the subject to them, but that don’t obviously or straightforwardly represent those external conditions.Footnote 47 In this light it makes little sense to insist that such experiences—again, think of pain—better reveal the nature of (aspects of) the external world than their own experiential natures. Here claims of transparency appear weak. And it seems just as strained to insist that our conceptions of those experiences hide or massively distort their natures. Here the paper’s argument that self-presentation guarantees near perfect self-revelation appears strong. After all, does our conception of pain really entitle us to say that the experience of pain—pain as we feel it—has no intrinsic features?
The physicalist can nevertheless tough it out. She can insist that our conceptions of experience do hide or massively distort its nature and that any contrary arguments beg the question. She doesn’t thereby contradict herself. But this paper has shown, it hopes, how forced and ad hoc this move is, how little drives it but the will to uphold physicalism come what may.
3.5 A Posteriori Physicalism Misses the Mark
Most a posteriori physicalists purport to explain why our phenomenal concepts fail to represent experience as physical. But they neglect to explain away the fact that we both experience and represent experience as having features (at least apparently) over and above the physical. We experience and represent it, that is, not merely as not physical, but also as (at least apparently) non-physical—as having intrinsic non-causal/functional properties that physical entities, understood in causal/functional terms, can’t have. So most a posteriori strategies, at the end of the day, fail to address the full problem of consciousness.Footnote 48 We can bring this out vividly. Let feature X be the intrinsic feature (or features) of experience (at least apparently) over and above the physical—the feature that underlies the explanatory gap. The a posteriori physicalist admits he can’t explain how merely physical entities come together to yield feature X. So he turns instead to our concept of X. But this concept includes a representation as of X. The a posteriori physicalist’s gambit therefore succeeds only if he can physically explain this representation. But how can physical entities yield a representation as of X when they can’t yield X itself? The difficulty stands out starkly when we recall two things: (1) that experience doesn’t support a distinction between appearance and reality; and (2) that self-presentation entails substantial, perhaps near perfect, self-revelation. It follows that the representation as of X (as provided by the phenomenal concept) and X itself (as given in consciousness) are either the same thing or very nearly so, and certainly the same kind of thing. The physicalist therefore physically explains the phenomenal concept only if he physically explains consciousness itself. But, if he could do that, he wouldn’t have bothered with phenomenal concepts in the first place.Footnote 49
Here, again, a close look at the implied metaphysics uncovers problems with the physicalist’s epistemology.
4. The Third Premise: Non-reducibility of Intrinsic Nature
Section 3 establishes that experience has or involves intrinsic nature, or at least that the physicalist gives us no good reason not to take at face value the appearance that it does. The physicalist must therefore show that this intrinsic nature in fact reduces to purely relational or extrinsic nature—that each intrinsic property of experience is (at least token) identical to purely relational properties.
Daniel Stoljar begins a case to this effect.
[O]ne can derive the instantiation of an intrinsic property from a relational one just by shifting what thing you are talking about. For example, being a husband is a relational property of Jack Spratt, and being a wife is a relational property of his wife. But being married is an intrinsic property of the pair (or the sum) of Jack Spratt and his wife. To take a different example, it seems plausible to say that I have the property of having a hand intrinsically, but my having this property obviously follows from a relation between my hand and the rest of my body, and that the truth concerning this is a relational truth.Footnote 50
Pereboom resists this reasoning by distinguishing absolutely purely intrinsic properties from comparatively intrinsic properties.Footnote 51 Whereas each of an entity’s comparatively intrinsic properties reduces to extrinsic properties of that entity’s parts, none of an entity’s absolutely purely intrinsic properties does.Footnote 52 Properties like having a given shape, having a hand and being a married couple count as comparatively intrinsic properties of their instantiating entities. But which kind of property is consciousness?Footnote 53
Maybe a quick answer lies at hand? After all, we seem to contradict ourselves when we deny that consciousness counts as, or at least involves, an absolutely purely intrinsic property. Recall that section 2 defined ‘intrinsic properties’ to cover properties other than spatio-temporal (structural) properties, dispositional (dynamical and/or causal/functional) properties, and all other relational properties. Section 3 then showed that conscious properties have this kind of intrinsic nature. And section 2 showed that physical entities, as physicalism must understand them, don’t and can’t. The paper thus framed its definitions so as to make it hard to see how the intrinsic nature of consciousness could reduce to the extrinsic nature of physical reality. (This result doesn’t issue from definitional sleight of hand. It took honest work to conclude that consciousness and physical nature satisfy the respective definitions.) We see the contradiction, more clearly perhaps, if we set aside intrinsic and extrinsic properties and attend instead to the absurdity of identifying purely non-causal/functional properties (which characterize consciousness) with purely causal/functional properties (which exhaust physical nature).
The physicalist might still argue that consciousness, while admittedly apparently absolutely purely intrinsic, nevertheless in fact reduces to extrinsic (and hence physical) elements. But we saw in section 2 that basic physical dispositions, combined, merely yield higher-level dispositional structures, and that causal dynamics, combined, merely yield more of the same. We might then ask how intrinsic nature breaks down into extrinsic elements if combinations of extrinsic elements seem incapable of resulting in intrinsic nature.
The physicalist might dig in his heels. He might argue that intrinsic nature does so reduce even though we don’t see how, even though we don’t see how it ever could. But here he makes the same kinds of epistemological moves treated in section 3. More, these moves, because they apply to conscious and physical entities just like their section 3 analogues, presuppose the same kinds of underlying metaphysics. The same conclusion therefore holds: we have good metaphysical reasons to dismiss these moves.
The physicalist might dig in his heels even more resolutely. He might deny the existence of intrinsic or qualitative nature altogether, as do those who embrace a purely power-based or dispositional view of reality (dispositions all the way down) and those extreme structuralists who understand nature as structures ungrounded in any intrinsic or qualitative reality (free-standing structures not instantiated by anything—dances without dancers).Footnote 54 This isn’t the place to argue at length against the powers conception or extreme metaphysical structuralism. Suffice it to say, however, that each such metaphysic, by denying intrinsic, qualitative nature, either denies consciousness outright, or else diagnoses it as an illusion, as a misrepresentation of purely dispositional structure. But section 3 showed that consciousness can’t sustain a distinction between reality and appearance. This rules out consciousness as an illusion of something else. The powers-view and extreme structuralism therefore imply, at the end of the day, the elimination of consciousness. Again, this is no place to tackle eliminativism about consciousness.Footnote 55 The paper instead takes the existence of experience for granted and treats its denial as a reductio of any doctrine with elimination as a consequence.
Where does this leave us? On its face, it seems implausible that intrinsic nature could reduce to purely extrinsic nature. It seems implausible that qualities, at bottom, are nothing but relations. It seems implausible that the world contains structure and nothing but. The physicalist who argues otherwise must tread one of two paths. He must make epistemological claims whose metaphysical presuppositions section 3 saw fit to reject. Or he must endorse a metaphysics implying the non-existence of consciousness itself. Yet it seems that we have far better reason to believe in consciousness than in any philosophical doctrine, much less one incompatible with consciousness. Indeed, if, as the paper suggests, we, as subjects, consist of our conscious experiences, then we can have no good reason to entertain a theory which denies the existence of consciousness.
5. Conclusion
Section 1 set out the paper’s metaphysical argument against physicalism. Sections 2 through 4 gave support, respectively, to each of the argument’s three premises. These sections sought, wherever possible, to adduce metaphysical rather than merely epistemological considerations. In this the paper departs from standard anti-physicalist approaches current in the literature. Now, if this support holds up, then the argument’s sub-conclusion, that conscious experience has properties beyond the physical, follows straightforwardly. As does the conclusion, that physicalism is false. This still leaves open which metaphysical theory does the world justice where physicalism doesn’t. The paper, through its arguments and sub-conclusions, hopes, however, to have laid down some metaphysical constraints on any such theory.
