Russellian Panpsychism: Too Good to Be True?

American Philosophical Quarterly Volume 52, Number 1, January 2015 RUSSELLIAN PANPSYCHISM: TOO GOOD TO BE TRUE? Patrick Kuehner Lewtas abstract Russellian panpsychism puts basic conscious properties at the bottom level and then grounds lowestlevel physical entities in them. This paper offers arguments against the view. (1) The explanatory gap cuts both ways, making it as hard to get the physical out of consciousness as to get consciousness out of the physical. (2) Russellian panpsychism can’t explain how basic conscious properties yield high-level consciousness. (3) Other non-physicalist views can evade the causal argument for physicalism at least as well as Russellian panpsychism. (4) Simplicity and beauty don’t supply reasons for Russellian panpsychism. R ussellian panpsychism arises from “mind-dust” panpsychism. Mind-dust panpsychism bestows basic conscious properties on basic physical objects.1 It then builds complex conscious properties—minds like yours and mine—out of these basic conscious properties. Mind-dust panpsychism thus (1) understands consciousness as a property; (2) reduces high-level consciousness to bottom-level consciousness; (3) declines to reduce bottom-level consciousness to anything non-conscious; and (4) accepts the existence of the irreducibly non-conscious physical. The mind-dust panpsychist needn’t claim that all basic physical objects have conscious properties. He only has to put enough basic consciousness “down there” to build high-level consciousness “up here.” So he might say that quarks, but not electrons, are conscious. Nor need he insist that all high-level objects are conscious. His panpsychism must include an account of what Nagel (1979) calls “mental chemistry”—the processes by which ©2015 by the Board of Trustees of the University of Illinois lower-level consciousnesses yield higherlevel consciousness. This account might tie mental chemistry to physical structure. The panpsychist could then say that brains, but not trees, have the kind of physical structure that allows many “little” consciousnesses to merge into one “big” consciousness. Now, Russellian panpsychism goes along with the first three of the above points but not the fourth. It instead grounds lowestlevel physical properties, understood as dispositions to interact thus-and-so with other such dispositions, in basic conscious properties. It thus rids its metaphysic of irreducibly non-experiential concrete entities. We should regard Russellian panpsychism as mind-dust panpsychism plus something more. But the differences run deep. Whereas mind-dust panpsychism puts forward a kind of property dualism, Russellian panpsychism offers something akin to idealist monism.2 Nowadays, most panpsychists opt for Russellian panpsychism.3 This paper urges them to think again. 58 / AMERICAN PHILOSOPHICAL QUARTERLY I. Panpsychism issues from a steadfastly rationalist outlook rooted in a strongish principle of sufficient reason. The view therefore spurns metaphysical and epistemological gaps: it insists that all high-level entities reduce, and reduce intelligibly, to bottom-level entities. It parts with radical emergentism because emergentism allows high-level entities that don’t reduce to bottom-level entities.4 It parts with physicalism because physicalism allows high-level entities that reduce, but not intelligibly, to bottom-level entities. 5 Because panpsychism holds metaphysics and epistemology to the same reductive standard, it keeps metaphysical and epistemological explanations from drifting apart. In each case, a good explanation shows how high-level entities are nothing over above basic bottomlevel entities. Panpsychism thus puts itself forth as an explanatory theory offering what it sees as the best explanation of high-level consciousness. Any panpsychist explanation must do two things. It must advance a theory of basic bottom-level consciousness. And it must present an account of mental chemistry. The theory of basic bottom-level consciousness must meet at least two constraints. It must paint a defensible picture of bottom-level consciousness. And it must provide the raw material for higher-level consciousness. Russellian panpsychism adds a third constraint: the basic conscious properties must also ground bottom-level physical dispositions. The theory of mental chemistry can take one of two forms. It can show how an arrangement of basic conscious properties constitutes a higher-level conscious property, the way a pile of blocks constitutes a tower. This kind of mental chemistry rests on synchronic supervenience. Chalmers (2012) labels it “constitutive” because the higher-level consciousness just is the collection of lower-level consciousnesses. Or the theory of mental chemistry can show how the basic conscious properties serve as raw material for something different from them, yet still nothing over and above them—the way red and yellow paint serve as raw material for orange paint. This kind of mental chemistry rests on diachronic causation. Chalmers labels it “nonconstitutive” because the higher-level consciousness doesn’t consist of the collection of lower-level consciousnesses, which may change, or even lose their individual identities, when forged into a single higher-level consciousness. In this context and others, the paper uses the results relation—results from, results in—to encompass the ways a higher-level entity can be nothing over and above one or more lowerlevel entities. The results relation covers relations like identity, composition, realization, and so on. Panpsychist rationalism entails two general constraints. First, the panpsychist can’t let emergence into his theory. After all, if anything emerges, why not consciousness? Allowing even the possibility of emergence thus shreds panpsychism’s warrant as against emergentism about consciousness. Second, the panpsychist can’t abide explanatory gaps.6 If panpsychism posits bottom-level consciousness to explain high-level consciousness, but has explanatory gaps, then it fares no better than physicalism. Because physicalism has greater prima facie plausibility—on grounds of simplicity, and so forth—we should prefer panpsychism only if it supplies a better explanation. Thus explanatory gaps undercut panpsychism’s warrant as against physicalism. The Russellian panpsychist accepts these constraints because they help justify his panpsychism. Further considerations then lead him to Russellian panpsychism. First, he finds it natural that categorical conscious properties ground lowest-level physical dispositions.7 Both he and his mind-dust friends understand physical nature as dispositions to interact thus-and-so with other such disposi- RUSSELLIAN PANPSYCHISM / tions (see section III). His rationalism then urges that categorical grounds prop these dispositions up. For otherwise the dispositions float free, not only had by nothing, but also powers to affect nothing more than other dispositions themselves had by nothing. And this, he worries, strips physical dispositions of all metaphysical heft. But which categorical grounds? Here, the Russellian notes that only conscious properties, of the concrete entities we know, have the non-dispositional nature categorical grounds need. All other known concrete properties, such as microphysical properties, seem dispositional upon closer look. And all other known categorical properties, such as shape, seem quasi-mathematical rather than unambiguously concrete, and therefore perhaps in need of underlying concrete grounds themselves. The Russellian has found a lock that needs a key. He has found a key that fits the lock. And this key is the only key he knows. The conclusion that the key belongs to the lock strikes him as too good to be false.8 Second, the Russellian finds his view lovely. It cleans the world of loose ends and brute links (no more “it just so happens that basic physical objects have basic conscious properties”). It solves in one go both the mind-body problem and the problem of the intrinsic nature of the physical. And it leads to a pleasing monism where most nonphysicalist ontologies end at ugly dualism. Third, the Russellian believes that only his theory allows for non-physical yet causally relevant conscious properties. Here he has his eye on the causal argument for physicalism. This argument token identifies conscious properties with physical properties on the grounds that (1) conscious properties cause physical events (understood, for these purposes, as physical properties instanced at times),9 (2) all physical events have sufficient physical causes (the doctrine of physical causal closure), and (3) systematic over-determination doesn’t occur.10 Non-physicalists don’t draw 59 the physicalist conclusion, of course. But most accept the second and third premises.11 This sets the price of non-physicalism at epiphenomenalism. And here Russellian panpsychism holds out hope of non-physicalism without epiphenomenalism. Maybe conscious properties can’t flex muscle in their own right. But if basic conscious properties give being and capacity to physical properties that can, then basic conscious properties put a causal stamp on the world nonetheless. Causation springs from them and would be nothing without them. And if higher-level conscious properties inherit causal relevance from the basic conscious properties that build them, they too make a difference.12 Russellian panpsychism puts constraints on the natures of dispositions and their grounds. It demands that dispositions have grounds, of course, while leaving open whether grounding takes the form of identity, realization, or something else. And it insists that each disposition result intelligibly from its grounds. For otherwise the Russellian ushers explanatory gaps into his theory. This not only runs counter to his rationalism, but also weakens his case against physicalism, perhaps fatally. The Russellian therefore can’t stomach grounding relations as radically contingent as those envisaged by Armstrong (1997) and Lewis (2009). These thinkers see no necessary link between the grounding property and the disposition(s) it supports. In fact, any suitably intrinsic property could ground any disposition given the right laws of nature. This means that grounding requires not only the so-called grounding property, but also a basic law of nature ontologically independent of it. And this infects the grounding relation with emergence. Nor can the Russellian accept, unqualified, Heil’s (2003) identification of dispositions with their grounds. The Russellian will refuse to identify properties lacking intelligible connections between them—properties, that is, separated by explanatory gaps.13 60 / AMERICAN PHILOSOPHICAL QUARTERLY What does it take to ground intelligibly so as to avoid explanatory gaps? The Russellian would seem driven to a necessitation relation understandable in terms involving some kind of a priori entailment.14 Otherwise, the ground would not seem to necessitate the disposition, which, for its part, would not seem to issue solely from the ground. An explanatory gap would lie between them. (Note that the entailment base can, and indeed must, include laws of nature that codify powers and susceptibilities intrinsic to the natures of the relevant entities. For these “laws” merely express our grasp of those natures. The base must nevertheless exclude laws of nature over and above such natures. Otherwise, the grounding relation features emergence.) This paper speaks of intelligible resulting rather than a priori entailing because it deals with the metaphysics of Russellian panpsychism rather than its epistemology. Although our concept of intelligible resulting may not be purely metaphysical, our concept of a priori entailment is altogether epistemological. Note that panpsychist rationalism, minddust or Russellian, rules out Heil’s (2003) identification of dispositions with the entities that have them, and his identification of intrinsic qualities with causal powers, at least where the identifications fly in the face of explanatory gaps. The panpsychist therefore can’t accept Heil’s identification of conscious qualities with physical dispositions/powers. II. This article offers arguments against Russellian panpsychism. Some attack the view itself. Others undermine the goals and drives behind it. We begin with explanatory gaps. The paper maintains that we can no more see how to “get” non-experience out of experience than we can see how to “get” experience out of non-experience. The next section refines the argument by shifting from non-experience to the more determinate physical. But we can draw useful conclusions even at the general level of experience and non-experience. These conclusions add to rather than repeat those reached later. The explanatory gap between consciousness and the physical—and thus, by extension, between experience and non-experience (since our conceptions of determinables can’t be more determinate than our conceptions of their determinates)—has two parts. First, we don’t see how to get consciousness out of the physical. Second, we don’t see even the possibility of getting consciousness out of the physical. So the explanatory gap does more than mark our current ignorance. It also points to an apparent impossibility. Now, why should the Russellian think we can get non-experience out of experience when we can’t get experience out of nonexperience? Put otherwise, if the Russellian thinks experience has enough in common with non-experience to ground it, how can he nevertheless insist that experience couldn’t result from non-experience? He can’t argue that experience involves intrinsic and/or categorical nature whereas non-experience, as such, doesn’t, because at this point—at this level of abstraction—we have no reason to think non-experience doesn’t involve intrinsic and/or categorical nature. At this level of abstraction, we know nothing about non-experience beyond its lacking experience. The Russellian must therefore base his denial that experience results from nonexperience only on the fact that experience and non-experience look so different. He has nothing else to go on. But, at this level of abstraction, we have no reason to think this difference works asymmetrically and every reason to think it doesn’t. The Russellian thus finds himself in a bad spot. Any claim that the difference works asymmetrically would seem ad hoc. And any argument bridging the difference would appear to cut both ways, undercutting the Russellian’s rejection of physicalism. RUSSELLIAN PANPSYCHISM / III. Nothing about experience, as such, even so much as hints at how physical nature might issue from it. But we can say more. Most panpsychists share understandings of consciousness and physical nature. They take their cue for the latter from science. They note that physics attributes two kinds of properties to basic physical entities: spatio-temporal properties (e.g., location) and causal dispositions (e.g., electric charge). Thus we learn that an electron is a spatio-temporal point or region with sundry dispositions to interact causally with other similarly located and causally disposed points or regions. Physics can’t say more. Science can only gather data about manifested dispositions—because all data must causally affect scientists’ senses (either directly, or indirectly through prior effects on instruments)15—and then explain them by postulating entities suitably disposed to cause those data. Hypotheses framed in terms of non-dispositional entities would lack explanatory power and hold aloof from empirical test. Science thus remains silent about any intrinsic or non-dispositional properties that might fill out physical structure, ground physical dispositions, or characterize physical nature apart from its relations to other things. Science then builds higher-level entities out of these basic and bottom-level dispositionclusters. Thus physical reality, as revealed by science, consists solely of spatio-temporally located dispositions and arrangements of them. The Russellian argues for intrinsic entities that ground these physical dispositions (because ungrounded dispositions are as flimsy as the Cheshire Cat’s grin without the Cheshire Cat) and that supply them with nonrelational relata (because, at the end of the day, relations merely relating other relations, then looping back on themselves, relate nothing and amount to nothing). But then he puts forward a metaphysical thesis going beyond 61 not only the best current science but also the best possible science. We can, of course, define physical nature any way we want, and could stipulate that it includes these intrinsic properties. But it will keep things cleaner to understand physical nature as physics and science must give it to us. Among other things, we otherwise understand it in ways physicalists can’t, and thereby garble the contrast between physicalism and panpsychism.16 Panpsychists understand conscious properties as having intrinsic and non-dispositional nature. Here, “intrinsic properties” means the properties of a concrete entity other than (1) its spatio-temporal properties; (2) its dispositional properties; and (3) any relations it bears to other entities not part of itself (including contrived relations).17 Intrinsic properties cover the sorts of things about which science says nothing—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. And “nondispositional property” means any concrete property other than a dispositional property. So panpsychists think conscious properties have intrinsic properties over and above whatever causal dispositions and relations they might also have. Panpsychists believe this because they directly experience these properties. But their view also forces them to see things this way. If they decided that causal dispositions exhaust consciousness, they would surrender any reason to reject physicalism. Russellian panpsychism, in particular, finds itself committed to these understandings. It must accord consciousness an essentially non-causal core to ward off the threat of physicalism. It also has no choice about a thoroughly dispositional, hence essentially causal, physical. Otherwise, it couldn’t call for categorical grounds. 62 / AMERICAN PHILOSOPHICAL QUARTERLY We now have a better handle on the consciousness/physical explanatory gap than we ever could on the experience/non-experience gap. We see that conscious properties have features physical properties don’t and can’t— at least as science reveals them. We thus better understand why we don’t see how to get consciousness out of the physical. More, we better understand why we don’t see how we ever could. Could conscious properties serve as the grounds underlying physical nature as understood here—granting, for now, the need for grounds of some kind? It seems they couldn’t. To see why not, we look at two cases. First, suppose conscious properties are entirely non-causal/non-dispositional.18 Their natures don’t include capacities to flex muscle and thereby make things happen. Note that this needn’t entail epiphenomenalism. Passive conscious properties could still make a difference if other entities—physical entities—actively responded to them.19 But if conscious properties are altogether passive, they couldn’t ground, so as to give being to, the dispositions exhausting physical nature. Active powers can’t issue from the absence of any and all active powers. Certainly to claim otherwise saddles panpsychism with an explanatory gap, flies in the face of its rationalism, and destroys its case against physicalism. We can’t evade this result by downgrading physical properties to passive entities other physical entities respond to, the way we can save passive conscious properties from epiphenomenalism. Even this dodge calls for at least some physical properties with capacities to respond actively. Otherwise, no responding happens and causation doesn’t occur. But these surviving capacities must, on the Russellian model, flow from their conscious grounds. Here, we face the selfsame problem, bounded, but just as bad. The second case treats conscious properties that do have a causal/dispositional side—as though grafted onto their intrinsic phenomenal side. This seems, at first blush, to give the Russellian what he needs. For now the conscious property’s causal/dispositional side can ground the physical property, allowing physical powers to issue from causal efficacy rather than causal impotence. But first blush crumbles before sober second look. For the Russellian grounds the physical property in something other than the conscious property’s intrinsic phenomenal nature. He therefore doesn’t ground it in consciousness at all. He grounds it instead in something else, something dispositional. And this, at the end of the day, leaves the physical property ungrounded, because dispositional grounds themselves need grounds. Perhaps the Russellian pleads that only a conceptual distinction marks off the conscious property’s phenomenal side from its causal/dispositional side. He claims, in other words, that the conscious property exists as an indivisible unity that we humans, put together as we are, can in thought cut into two. But this reply fails. The Russellian thereby commits himself to an explanatory gap, and maybe an absurdity, since we don’t and can’t see how the essentially causal can equal the essentially non-causal. Besides, if the Russellian admits that non-causal nature can equal causal nature even though we don’t and can’t see how, then he lets metaphysics and epistemology pull apart and gives the game to the physicalist.20 Note that these considerations doom the Russellian claim to have secured, alone among nonphysicalisms, a place for causally effective consciousness. Not only can’t Russellian panpsychism deliver the promised goods, but also, as remarked above, property dualisms, like emergentism and mind-dust panpsychism, can accord passive conscious properties causal efficacy by giving physical entities the capacity to respond actively to them. This removes perhaps the most influential motivation for Russellian panpsychism.21 RUSSELLIAN PANPSYCHISM / IV. Mind-dust basic conscious properties must satisfy two sets of constraints. They must make possible a believable theory of bottom-level consciousness. And they must enable mental chemistry. These constraints won’t work at cross purposes because they target different aspects of bottom-level consciousness. The first set addresses the form of basic conscious properties: for example, how simple they are, whether they respond to outside influences, how many a basic physical object has, and so forth. The second set deals with their phenomenal contents: for example, which experiences characterize basic physical objects, whether some experiences (e.g., orange-experiences) result from more basic experiences (e.g., red-experiences and yellow-experiences), and so on. It doesn’t go so easily with Russellian basic conscious properties, however. Because these ground the physical, they must also satisfy a third set of constraints.22 These constraints touch upon phenomenal form and content, like the others, but tug in different directions. Elsewhere, the author puts forward theories of mind-dust basic consciousness and mind-dust mental chemistry.23 These build upon constraints that force either unique theories or at least much narrowed ranges of theories. Two results have especial relevance here. One states that basic conscious properties are simple—either maximally simple and wholly unstructured or at least extremely simple and only minimally structured. This follows from a principle tying complexity of experience to complexity of physical structure. (This makes sense given the dependence of mental chemistry on physical structure—see section V.) For instance, mice surely have simpler experiences than ours given their simpler brains. By the same token, most assume that quarkexperiences, if real, are rudimentary.24 The second result notes the many kinds of basic 63 conscious properties—many more than the kinds of basic physical objects to instantiate them or the kinds of basic physical properties to match them up with. This awkward result follows from the need to put “down there” everything it takes to build you, me, and bats “up here.” The bottom level must thus have basic forms of any type of experience basic with respect to all other types of experience. Here we treat one type of experience A as basic with respect to another type of experience B when (1) we don’t, and don’t see that we could, see how instances of A could result from instances of B; and (2) we don’t, and don’t see that we could, see how instances of A and B could both result from experience(s) of the same type(s) C. (We can use the explanatory gap as a test for basic-ness because of the way panpsychism links metaphysics and epistemology.) Now, any experience from any sensory modality (e.g., red-experience) seems basic with respect to any experience from another sensory modality (e.g., salty-taste-experience). We must bear in mind here sensory modalities beyond humans but found in other species as well as those not expressed in any species but nevertheless possible.25 Within each sensory modality we have sundry mutually basic experience-types (e.g., the three basic color experience-types, the five basic taste experience-types, the several thousand basic smell experience-types, and so forth). We also have the experience-types from which emotions result—human emotions, animal emotions, and emotions as alien to us as guilt is to lizards.26 Nor should we forget the basic experience-types, if any, underlying occurrent attitudes, thoughts, understandings, and so on. The Russellian grounding constraints clash with these results in two ways. First, they lead to a mismatch of number. The grounding constraints limit the number of distinct basic conscious properties to the number of distinct basic physical entities there to ground—two 64 / AMERICAN PHILOSOPHICAL QUARTERLY or three dozen at most, only a handful of which serve as building blocks of terrestrial organisms. But this number falls far short of the crowd of distinct basic conscious properties entailed by the other constraints.27 The Russellian can’t make ends meet here by paring down his stock of distinct basic conscious properties. For he then builds erstwhile basic conscious properties in explanatory-gapdefying ways (e.g., “building” salty-tasteexperiences out of basic red-experiences and basic anger-experiences). (The Russellian might also resort to a many-to-one grounding relation, where each of many distinct basic conscious properties grounds a given basic physical entity. Section V explains why this won’t do.) The grounding constraints also lead to a mismatch of structure. They entail that the structures of the grounding properties match the structures of the grounded properties as needed for the former to result in the latter. We can envision two kinds of mismatch here: a mismatch of complexity (of degree) and a mismatch of kind. The mismatch of complexity arises from the extreme, even maximal, simplicity of basic conscious properties compared with the greater complexity of basic physical entities. Thus a basic red-experience might involve a homogeneous experience of red, a red without texture, variety, borders, relations, or change. But even basic physical entities have more complexity than this. Basic mass, for instance, has inertial and gravitational aspects (unified by the theory of general relativity), the second of which obeys an inverse square law.28 Basic charge comes in two forms, positive and negative, which interact differently and each obey an inverse square law. Basic spin has two opposite values that orient themselves along any spatial axis. And so on. Not easily do we extrude the architecture of these basic physical entities from the blobbiness of their alleged grounds. The relevant structures likely exhibit a mismatch of kind even if basic conscious properties have more complexity than admitted above. Three considerations seem apt. First, we can’t grasp how a basic conscious property could have a structure suitably isomorphic to an inverse square law, much less to an entity with two differently interacting forms each obeying an inverse square law. Second, nobody has proposed a match-up between phenomenal structure and physical structure, even though such a match-up would argue powerfully for (at least token) mindbody identities—much more powerfully than do mind-body correlations. And third, the grain problem points to the jaggedness of micro-physical structure compared to the smoothness of phenomenal structure.29 True, the phenomenology in question sits at a high level. But we can’t readily imagine basic conscious properties jagged enough to ground basic physical entities yet smooth enough to build smooth high-level experiences.30 The Russellian might counter these arguments by positing complex and structured basic conscious properties. (He would have to outflank the arguments for simple and unstructured basic conscious properties, but let this pass for now.) One panpsychist has suggested a single basic conscious property—an ur-experience made of all possible basic experience-types much the way white light contains all colors of the spectrum.31 Here mental chemistry subtracts experiences instead of combining them. This paper finds the proposal unworkable. If the ur-experience is indeed made of other more basic experiences, then it doesn’t qualify as a basic conscious property, and we end up back where we started. On the other hand, if it does so qualify, then it seems unintelligible, maybe even impossible. We can’t see how incompatible and/or incommensurable experience-types could coexist in an undifferentiated whole. What undifferentiated whole could we get, after all, from joy-experience, salty-taste experience, and pain-experience (and try to imagine a whole, not a mosaic or succession)? We don’t readily see how subtraction could work RUSSELLIAN PANPSYCHISM / (especially when applied to an undifferentiated whole rather than a mosaic or succession). We have trouble even making sense of subtraction in the context of a basic property that, as basic, can’t have parts and can’t break down into smaller bits. Another panpsychist has ventured that each bottom-level experience exists as a structured composite of basic conscious properties.32 But we face difficulties here too. First, we have unacceptably complex bottom-level conscious properties, even if we allow the same complexity as bottom-level physical properties. We have to cram into four or five (at most a couple dozen) bottom-level conscious properties all the basic experience-types needed to build higher-level consciousness. Second, with bottom-level conscious properties much more complex than bottom-level physical properties, we again face a mismatch of complexity, but this time the other way round. True, a complex entity can more easily ground a simpler entity than can a simple entity ground a more complex entity—because the grounding relation doesn’t have to create complexity out of nothing. But the grounding property still needs the right kind of complexity. Now, the grounding constraints call for a structure apt for grounding a basic physical property; and the other constraints call for a structure apt for building higher-level consciousness. We have no guarantee that a single structure could fulfill both job descriptions (and the grain problem counsels pessimism). Even if it could, we might wonder how natural it is for a single bottom-level entity to play such disparate roles. Third, bottom-level but composite (and therefore non-basic) properties seem odd. We might wonder how a basic physical object instantiates a composite property. We might also wonder how a bottom-level but composite experience results from basic conscious properties never individually instantiated by anything. This section’s arguments (and argument sketches) don’t strike knock-down blows. 65 They show, however, that a well-elaborated Russellian panpsychism won’t have the lean elegance its defenders suppose. On the contrary, it will hold together, if at all, only through fancy footwork and theoretical clutter. Russellian panpsychism thus underscores, not the naturalness of wedding the physical to consciousness, but the loose metaphysical fit between them. V. Most panpsychists agree that physical structure determines whether and how a physical composite works mental chemistry on the conscious properties of its parts.33 Now, mind-dust mental chemistry not only builds high-level experiences out of bottomlevel experiences. It also builds high-level subjects out of bottom-level subjects. In each case, many input entities result in a single output entity. Mind-dust mental chemistry thus works by integration and transformation rather than mere summation. It does this by physically integrating (in mental-chemically effective ways) the physical objects that instantiate the right conscious raw material. For example, some physical structure might integrate smaller physical objects, some of which, on their own, experience red, and some of which, on their own, experience yellow, such that the overall object has a single subject with a single experience of orange.34 Chalmers, recall, dubs this kind of mental chemistry “nonconstitutive.” Russellian mental chemistry works differently. Its bottom-level conscious properties not only build high-level experiences, but also ground physical properties. Because these physical properties continue to exist even as mental chemistry occurs, so, too, must their underlying grounds. This keeps mental chemistry from integrating and transforming— from crafting raw material into something else. It forces instead a “constitutive” mental chemistry where the bottom-level conscious properties, as such but all together, equal the 66 / AMERICAN PHILOSOPHICAL QUARTERLY higher-level experience (much as “pixels” of color equal a pointillist painting). This picture suffers from two problems. First, it doesn’t build anything, and thus doesn’t offer an account of mental chemistry. Each pixel of consciousness has its own subject and its own experience, in each case as cut off from other subjects and experiences as you and yours are from me and mine. We don’t end up with an overall subject and experience any more than when we line people up, put them in a room, or jam them together.35 Russellian panpsychism can get an overall subject and experience, while keeping in place the grounds supporting the physical, only through the creation of something in addition. But the creation of something in addition fails. First, it involves emergence, since the new subject and experience don’t result from the pixels, but stand apart from them, having merely been triggered by them. Second, the creation of something in addition violates a principle of conservation of consciousness according to which quantity of consciousness can neither be created nor destroyed. Any workable panpsychism must honor such a principle. Not only does it follow from anti-emergence; not only do other basic properties that serve as building blocks obey conservation laws; but also mental chemistry presupposes it—for otherwise, we lose the metaphysical contrast between building and emergence and therewith the means to keep panpsychism free of processes incompatible with it.36 Third, the overall subject would, it seems, have to take the form of a homunculus whose experience, although not built out of the underlying conscious pixels, results from a kind of perception of them much as our television experiences result from blurred perceptions of TV pixels. Not only does this fall short of constitution. But it also involves direct perception of other experiences, in a word, telepathy, not a pillar upon which the Russellian should happily rest his philosophy. Russellian mental chemistry suffers from a second problem. It doesn’t have enough conscious raw material “down there” to build you and me “up here.”37 We’ve seen that Russellian panpsychism has room only for a few basic conscious properties—only as many as the basic physical entities they ground. This means Russellian panpsychism can’t explain us—unless it claims, for instance, that red experiences result from salty-taste experiences plus feeling-of-disgust experiences. But then it depends on explanatory gaps, with all that that entails. The Russellian does have a couple of moves. Perhaps he can deal with both problems by positing structured bottom-level conscious properties, each with a plurality of basic conscious properties arranged thusand-so. We came across these “glue-balls” in the last section. They might allow for enough conscious raw material “down there.” They might also allow for non-constitutive (i.e., workable) mental chemistry. In this case, mental chemistry would choose the right glue-ball, then pick out of the structured tangle of basic conscious properties the one(s) to feed into higher-level consciousness. Each glue-ball might have enough structure and categorical stuffing, in excess of what it needs for grounding, to lose a few basic conscious properties without disrupting the grounding relation and, with it, the supervening physical.38 All this notwithstanding, the problems with composite bottom-level conscious properties, canvassed in the last section, remain as bad as ever. The Russellian might also posit a many-toone grounding relation, where each of many distinct basic conscious properties grounds a given physical entity. This move addresses the shortage of conscious raw material, but at the cost of making mental chemistry more complicated. Mental chemistry now can’t put the right basic conscious properties in place merely by putting the right lowestlevel physical entities in place, because RUSSELLIAN PANPSYCHISM / each lowest-level physical entity can sit atop various basic conscious properties. It seems that mental chemistry must directly place the right basic conscious properties, in which case it needs to “see through” the physical to the underlying conscious grounds. This opens up the possibility of identical high-level physical structures supporting different high-level conscious states. And this violates high-level mind-body supervenience—no high-level mental difference without a high-level physical difference—a principle the panpsychist should not lightly cast aside given the dependence of mental chemistry on physical structure. (This move also increases the threat of a mismatch of complexity. For basic conscious properties must now ground, not only everyday causal powers, but also powers to directly see and place basic conscious properties. This widens the gap between the complexity of the grounding properties and the complexity of the grounded properties.) Of course, even if the Russellian somehow overcomes these difficulties, his many-to-one grounding relation does nothing to solve the problems with constitutive summation. VI. We saw in section I that simplicity and beauty sway the Russellian panpsychist. But maybe too much. We best understand simplicity as a norm with roots in epistemological prudence rather than metaphysical fact. Nothing binds the world to simple form. True, natural processes unfold so as to reach the lowest available energy levels given nature’s rules of the game. But nothing binds nature to simple rules of the game. There is nothing notably simple, much less ideally simple, about the world as given by the standard model of physics or quantum mechanics, let alone string theory or quantum gravity. Simplicity bites only when we craft theories. It tells us to postulate as little as we can and still explain our data. Otherwise, our theories float free of 67 their data—their one link to the world—and become arbitrary and unwarranted. Because the norm urges conservatism only insofar as we still explain our data, it holds only other things being equal. We should always run with a more complicated theory that explains better. Scientific theories have thus grown more complicated as science has improved. Beauty is something else again. It hardly counts as a norm at all—epistemological or metaphysical—hand-waving remarks from scientific oracles notwithstanding. What beauty do we find in diseased bodies or earthquakes; in the ceiling set by the speed of light; in the many dimensions of string theory or the slew of elementary particles? Don’t forget that beauty springs from needs and wishes as much as eternal truths. The slug exults in dank, dark muck. The vulture finds nothing so lovely as a stinking corpse. Yes, perhaps our world would cater to human cravings were physical dispositions anchored in basic consciousness. But that neither makes it so nor gives us reason to believe it. Maybe panpsychist rationalism requires that physical dispositions rise forth from categorical grounds. But it doesn’t, by itself, require that basic consciousness do the grounding.39 Maybe we don’t know of any concrete categorical properties besides consciousness. But that doesn’t mean they don’t exist and don’t underlie the physical. Don’t forget the arguments in sections II and III which, if successful, show that consciousness couldn’t underlie the physical. The world would be simple if it did, as would our best theory of it, but simplicity only governs other things being equal. And here, not only don’t conscious grounds explain physical dispositions, but, worse, they never could. Panpsychist rationalism, pushed hard, leads not to simple beauty but to unseemly mess. For if physical dispositions need categorical grounds, but consciousness can’t serve, then something else, something nonconscious and non-physical, must sustain 68 / AMERICAN PHILOSOPHICAL QUARTERLY the physical world. This yields a bottom level with non-conscious and non-physical categorical entities that also instantiate basic conscious properties. These basic conscious properties build higher-level conscious properties. And the non-conscious non-physical we-know-not-what upholds the physical. Panpsychist rationalism, pushed hard, thus joins the drawbacks of neutral monism to the drawbacks of panpsychism. This article remains undecided about whether physical nature does call for underlying grounds. In the meantime, it urges a more modest conclusion. The explanatory gap cuts both ways. It just as surely keeps consciousness from grounding the physical as it keeps the physical from engulfing consciousness. We find ourselves in an irreducibly dual world—complex, ugly, but real. American University of Beirut NOTES 1. Throughout, “basic” means not built of anything. 2. Russellian panpsychism differs from Russellian monism, which reduces bottom-level physical entities to basic conscious properties or basic proto-phenomenal properties (non-physical and nonconscious properties that yield conscious properties when suitably combined). This paper only addresses Russellian panpsychism. 3. Russellian panpsychists include Chalmers (in some moods: see Chalmers 2012, 2013); Lockwood (1989); Rosenberg (2004); and Strawson (2006). Mind-dust panpsychists include Nagel (1979); Strawson (1994); and the author. 4. Emergentism, in this context, asserts the existence of high-level but nevertheless basic properties: high-level properties not built of anything, much less lower-level or pre-existing entities. 5. Physicalism, very roughly, claims that all concrete entities are either physical or nothing over and above the physical. A priori physicalists claim that consciousness does reduce intelligibly to nonconscious entities. Panpsychists deny this claim. From their standpoint, all physicalisms combine metaphysical reduction with epistemic unintelligibility. 6. An explanatory gap exists wherever a metaphysical explanation fails to yield an epistemological explanation. We don’t face an explanatory gap when chemists reduce water to H2O: the properties of H2O make it clear why water exists and has the properties it does. But, according to the panpsychist, we do face an explanatory gap when physicalists reduce consciousness to physical entities: the properties of physical entities don’t make it clear why consciousness exists or has the properties it does. 7. Throughout, “categorical” means non-dispositional. 8. Seager (2006), pp. 135–138, summarizes this thinking. Chalmers (1996), pp. 137–138, advances it as an argument for Russellian views. 9. Different understandings of events lead to different versions of the argument. These nuances don’t matter here. 10. Systematic over-determination occurs if every instance of every conscious cause is over-determined by some (and not necessarily the same kind of) physical cause. 11. See Jackson (1982); Chalmers (1996); and Kim (2005). 12. See Chalmers (1996), pp. 136–140; Chalmers (2012, 2013). Rosenberg (2004) explores a Whiteheadian version of Russellian panpsychism grounding physical causation in experience. RUSSELLIAN PANPSYCHISM / 69 13. Some objections to the paper’s arguments target the grounding relation—claiming either that it involves the kind of radical contingency treated above or that nomic/causal facts and basic conscious properties serve jointly as grounds (such that each plays a necessary role). But in every case, we can ask whether the grounding relation counts as a results relation. If so, the paper’s arguments apply. If not, the proposed grounding relation runs afoul of the constraints in the text. 14. See Goff (2009) for arguments that panpsychist mental chemistry requires (a metaphysical analogue of) a priori entailment. These arguments apply as forcefully to the Russellian grounding relation. 15. Might manifested dispositions have non-dispositional properties where dispositions themselves don’t? It doesn’t matter. Even if non-dispositional manifestings do occur, science could know nothing of them. The scientist can only learn about entities disposed to affect the senses. This last bottleneck winnows out any information about non-dispositional goings-on. 16. Strawson (2006) describes his panpsychism as a kind of physicalism and identifies himself as a physicalist. This lends his work an undeserved air of sophistry. Lewtas (2014) argues that physicalism can recognize as physical only the kinds of entities science describes. Otherwise it undercuts its warrant, which rests significantly on the success of science. 17. See Langton (2001) and Pereboom (2011) for discussions of intrinsic properties. 18. Fully non-causal/non-dispositional conscious properties would also lack active susceptibilities. This doesn’t matter here. 19. See Lewtas (forthcoming b) for a theory of mental causation along these lines. 20. Many physicalists defend their view through concept dualism (see the papers in Alter and Walter 2007). Concept dualism claims that our picture of the world has finer grain than the world itself. This allows for two distinct concepts with wholly independent criteria of application but the same referent. 21. Some Russellian-like views claim that basic conscious properties, with their intrinsic natures, give concrete reality to physical dispositions but don’t in fact result in them. Thus physical structures and basic consciousness have equal ontological standing and together yield the concrete world (think, roughly, of Aristotle’s form and matter). It remains unclear whether such views count as kinds of Russellian panpsychism (they don’t reduce the physical to the phenomenal). In any case, this paper sets them aside. Not only does it deem them implausible, but also it can’t judge them without appraising physical/scientific structuralism and Aristotelian-style metaphysics—tasks impossible to carry out here. 22. Chalmers (2012, 2013) accordingly observes that Russellian panpsychists theorize within very tight constraints. 23. Lewtas (2013, forthcoming a). 24. Thus Seager (1999), p. 245, describes bottom-level mentality as “very circumscribed and impoverished.” Such is the stuff dreams are made of! 25. Two examples: sea turtles “see” the Earth’s magnetic field; fish “feel” pressure changes with their lateral lines. 26. Wade and Tavris (2000), pp. 406–408, suggest that a handful of primary emotions serve as raw material for other secondary emotions. 27. Chalmers (2012) notes this “large palette problem” as a challenge for Russellian panspychists. He hopes they find a “small palette” of basic conscious properties few enough to ground bottom-level physical dispositions yet rich enough to result in higher-level consciousness. He admits the prospects look bleak. 70 / AMERICAN PHILOSOPHICAL QUARTERLY 28. This ignores the Higgs field, which, if real, demotes mass to non-basic status. The Higgs field, however, has more structure than basic mass, so the paper’s argument still goes through. 29. Sellars (1965) first formulated the grain problem. Chalmers (2012) notes the challenge it poses for Russellian panpsychism. 30. Chalmers (2012) frames the structural problem as follows (but here recast in this paper’s terms). Basic conscious properties need structures much like the structures of the bottom-level physical dispositions they ground. Higher-level conscious properties and higher-level physical properties both result, by construction, from these basic conscious properties. Then why do higher-level physical structure and higher-level phenomenal structure differ so? Chalmers distinguishes this problem from the grain problem, which, as he sees it, attends specifically to the issue of jaggedness versus smoothness. 31. Keith Turausky presented this idea at a July 2012 workshop in Australia. 32. Luke Roelofs (personal communication). 33. Not only does physical structure seem apt for the role, but nothing else—no other factor plausibly available—does. See Lewtas (forthcoming a) for more. 34. This must happen without emergence—without high-level structures with mental-chemical powers not built out of bottom-level powers. 35. Here we have the panpsychist “combination problem,” named by Seager (1999), most forcefully put by James (1890/1950) (who cast it in the terms used in the text), and discussed by all panpsychists. Goff (2009) and Coleman (2014) argue that no plurality of subjects logically entails, or metaphysically necessitates, the existence of another subject. Their arguments strike hard at Russellian panpsychism but, in the author’s view, bounce harmlessly off non-constitutive panpsychism. 36. Lewtas (forthcoming a) says more about this principle. 37. Again, Chalmers’s palette problem. 38. The glue-ball loses the conscious property fed into higher-level consciousness as a consequence of the conservation principle. Otherwise the higher-level consciousness emerges from, instead of resulting from, the property in the glue-ball. 39. In rhetorical answer to the Russellian’s “Why not phenomenal properties for [the grounding] role?,” Levine (2001), p. 25, asks: “Well, why yes?” REFERENCES Alter, Torin, and Sven Walter. 2007. Phenomenal Concepts and Phenomenal Knowledge (New York: Oxford University Press). Armstrong, D. M. 1997. A World of States of Affairs (Cambridge, UK: Cambridge University Press). Chalmers, David J. 1996. The Conscious Mind: In Search of a Fundamental Theory (New York: Oxford University Press). ———. 2012. “The Combination Problem for Panpsychism.” http://consc.net/papers/combination.pdf. ———. 2013. “Panpsychism and Panprotopsychism,” The Amherst Lecture in Philosophy, Lecture 8. http://www.amherstlecture.org/chalmers2013/. Coleman, Sam. 2014. “The Real Combination Problem: Panpsychism, Micro-Subjects, and Emergence,” Erkenntnis, vol. 79, no. 1, pp. 19–44. Goff, Philip. 2009. “Why Panpsychism Doesn’t Help Us Explain Consciousness,” Dialectica, vol. 63, no. 3, pp. 289–311. Heil, John. 2003. From an Ontological Point of View (New York: Oxford University Press). RUSSELLIAN PANPSYCHISM / 71 Jackson, Frank. 1982. “Epiphenomenal Qualia,” Philosophical Quarterly, vol. 32, no. 127, pp. 127–136. James, William. 1890. Principles of Psychology (Repr., New York: Dover Publications, 1950). Kim, Jaegwon. 2005. Physicalism, Or Something Near Enough (Princeton, NJ: Princeton University Press). Langton, Rae. 2001. Kantian Humility: Our Ignorance of Things in Themselves (New York: Oxford University Press). Levine, Joseph. 2001. Purple Haze: The Puzzle of Consciousness (New York: Oxford University Press). Lewis, David. 2009. “Ramseyan Humility,” in Conceptual Analysis and Philosophical Naturalism, ed. David Braddon-Mitchell and Robert Nola (Cambridge, MA: MIT Press), pp. 203–222. Lewtas, Patrick. 2013. “What It Is Like to Be a Quark,” Journal of Consciousness Studies, vol. 20, nos. 9–10, pp. 39–64. ———. 2014. “Physicalism and the Intrinsic Nature of Consciousness,” Dialogue, vol. 53, no. 2, pp. 203–228. ———. Forthcoming a. “Building Minds: Solving the Combination Problem.” Unpublished manuscript. ———. Forthcoming b. “Passive Causation: Making Interactionism Work.” Unpublished manuscript. Lockwood, Michael. 1989. Mind, Brain, and the Quantum (Oxford, UK: Blackwell). Nagel, Thomas. 1979. “Panpsychism,” in Mortal Questions (New York: Cambridge University Press), pp. 181–195. Pereboom, Derk. 2011. Consciousness and the Prospects of Physicalism (New York: Oxford University Press). Rosenberg, Gregg. 2004. A Place for Consciousness: Probing the Deep Structure of the Natural World (New York: Oxford University Press). Seager, William. 1999. Theories of Consciousness (New York: Routledge). ———. 2006. “The ‘Intrinsic Nature’ Argument for Panpsychism,” in Consciousness and Its Place in Nature, by Galen Strawson et al., ed. Anthony Freeman (Exeter, UK: Imprint Academic), pp. 129–145. Sellars, Wilfrid. 1965. “The Identity Approach to the Mind-Body Problem,” Review of Metaphysics, vol. 18, no. 3, pp. 430–451. Strawson, Galen. 1994. Mental Reality (Cambridge, MA: MIT Press). ———. 2006. “Realistic Monism: Why Physicalism Entails Panpsychism,” in Consciousness and Its Place in Nature, by Galen Strawson et al., ed. Anthony Freeman (Exeter, UK: Imprint Academic), pp. 3–31. Wade, Carole, and Carol Tavris. 2000. Psychology (6th edition) (Upper Saddle River, NJ: Prentice Hall).