89

According to the Canberra Plan, the first step in a reductive physicalist enterprise is to functionally define the property to be reduced, and the second step is to find the physical property that fills that functional role. Reductive physicalism is true for the mind if both steps can always be carried out for mental properties. This picture of what reductive physicalism is stems from J. ]. C. Smart’s (1959) “topic-neutral” analyses and has been advocated in one form or another by Armstrong, Chalmers, Jackson, Kim, Levine, and Lewis, even though these figures differ from one another on whether they are proponents or opponents of reductive physicalism. Smart’s r959 article is also the source of a different and perhaps incompatible picture, the mind-body identity view, according to which reductive physicalism about the mind should be modeled on “theoretical identities” such as light= electromagnetic radiation (of wavelength 400-700 nm). This chapter will argue thar the point of view of the Canberra Plan neglects ground. I will consider a few attempts to grafr an account of the physical/ functional ground of mind onto the Canberra Plan, arguing that such attempts lead nowhere. Terminological note: my main point is that reductive physicalism requires that, for any phenomenological similarity between two mental states, that similarity must hold in virtue of a physical similariry that explains or constitutes the phenomenal similarity; that is, the fact that there is a phenomenal similarity has as its ground the fact that there is a physical similariry. This terminology, in which variants of ground, in virtue of, explains (understood metaphysically rather than epistemically) are used to connect sentences, fits the way of speaking of ground in the work of Kit ‘ Fine. However, I also use an abbreviatory device not used in Fine’s papers .· in which I speak of a phenomenal similarity as grounded in a physical 105 106 NED BLOCK similarity. Thar is, I talk of properdes being grounded in other properties as well as of facts being grounded m other facts. A second item of terminology is that I use the terms second orde~ and first order not in the sense of properties of properties versus properties of individuals, bur as follows: a second-order property is a p~operty that has a true definition in terms of having some other properties that meet a certain sort of condition. And a first-order property does not have such a true definition. A functional property is a special kind of second-order property in which rhe definition specifies ca_usal relat1ons ~o oth:,r properties and to inputs and outputs. I am suppoSl:”g that the other properties quantified over in the second-order definmon are themselves first-order properties. II. The Canberra Plan I start by describing some of the views that neglect ground. . J Kim l’n his landmark works on mental causat10n and reduction aegwon, dlfd’ of mind (Kim, 1972, 1993a, l998b, 2005), argues for _a mo e _o re uct1ve physicalism as functional reduction. The first step m reducmg water to H O light to electromagnetic radiation of 400-700 nm, or the property of b:ing a gene to molecular aspects of DNA is to functionally define the property to be reduced. For ex~mple, the property of ~e~ng a gene might be defined in terms of its role m encoding and transm1trmg genetic · c · (Kim 199sb p 25· 2005 pp 101-2). The next step 1s to find the in1ormat1on , , . , , · reali2ers of the functional role that has been defined. For example, DNA molecules encode and transmit genetic information.1 The third sta~e is l · · th echan;om by which rhe realizer accomplishes that function, expammg em = . d’ d for example, how the DNA molecule actually does the )Ob of enco mg an transmitting. l h ( l · 1 ‘ Joe Levine (1993, p. 132) holds that: “Stage l invo ves t e re at1ve y. Quasi?) a priori process of workiug the concept of the property to . be reduced” into shape “for reduction by identifying the causal role for “‘:h.1ch we are seeking the underlying mechanisms. _Stage 2 mvolves the ,:,mpmcal work of discovering just what those underlymg mechamsms are. 1 ft n been ointed out, theories at the “upper level” are often incompacibl.ewith the th:ories ~ ~e;izers, sopthis description is lri?hly idealized. T~er~ is a ~oo: o;~e{:)~n~ literature about the consequence of this fact for reducnon1sm. ee ey z ~x::;’.’5~:r ;ea;;i:~;~::~: ~:~);eprinted in Block, Flanagan, and Gilzeldere (1997). These ideas are discussed in Block and Stalnaker (r999). The Canberra Plan neglects ground ro7 Frank Jackson (1994, l998b) tells a similar story. He sketches (1998b, p. 59) the following argument for the reduction of temperature to mean molecular kinetic energy: Pr.I [NB: premise I}: Temperature in gases = that which plays the temperature … role in gases. (Conceptual claim) Pr.2: That which plays the temperature role in gases = mean molecular kinetic energy. (Empirical discovery) Cone.: Temperature in gases = mean molecular kinetic energy. (Transitivity of’=’) Jackson notes that premise l, the conceptual analysis in functional terms, can be thought of in either of two ways: as a claim of synonymy or as capturing an a priori reference-fixing claim. These views share some crucial features with the “Canberra Plan” movement of J. J. C. Smart, David Armstrong, and David Lewis (1966, 1970, 1972). Lewis held that the meanings of mental state terms can be analyzed via definitions of the following form: the state with causal role R. If mental state M can be seen via a priori analysis to be the state with causal role R, and if brain state B is found empirically ro have causal role R, it follows that M = B. Lewis regarded supposed identities such as “pain = C-fiber stimulation” as contingent rather than necessary. The idea is that the term pain is a (nonrigid) definite description that picks out the contextually indicated property that occupies causal role R. One physical property might be picked out in the context of human pain, another in the context of octopus pain. Note also that Lewis regarded these identities as “typetype,” that is, as identifying the property pain with the property C-fiber stimulation, and not as identifying this particular pain with this particular instance of C-fiber stimulation.3 Since he held that (assuming brain state B has causal role R) M = B, Lewis is often said to be a physicalist -including by Lewis himself (prior to “Mad Pain and Martian Pain” [1980], which advocates a mixture of functionalism and physicalism). And since he accepted a priori causal role ‘ analyses of mental state terms, he is often considered a functionalist. Some

e the term fanctionalist for those who identify mental states with their causal roles, and on that definition Lewis is not a functionalist since he identified mental states with the realizers of those roles, not the roles themselves. David Chalmers (2012, p. 362) describes his view as “at least a close relative of the Canberra Plan,” even though he is skeptical of the explicit definition aspect of the view. However, he has in the past endorsed something that sounds very Canberrish, for example, here: For the most interesting phenomena that require explanation, including phenomena such as reproduction and learning, the relevant notions can usually be analyzed functionally. The core of such notions can be characterized in terms of the performance of some function or functions (where “function” is taken causally rather than teleologically), or in terms of the capacity to perform those functions. Ir follows that once we have explained how those functions are performed, then we have explained the phenomenon in question. Once we have explained how an organism performs the function of producing another organism, we have explained reproduction, for all it means to reproduce is to perform that function. (Chalmers, 1996, p. 43)4 Although Chalmers is skeptical about explicit definitions, his “scrutabiliry” framework shares the features of the Canberra Plan that I will be criticizing. The key similarity with the Canberra Plan is that reductive accounts are always accounts of determination by the reductive base without consideration of the ground of similarities in cases in which similar facts are determined by different reductive bases. In other words, Chalmers’s vision of the reductive physicalism that he rejects does not require that phenomenological similarities with different scrutability bases be explained by physical similarities in the scrutability bases. The Canberra Plan as I have been construing it is functionalist in that mental states are analyzed functionally in terms of their causal role. And it is physicalist in that mental states are said to be the physical occupants of these roles. Kim (1998b) has given what may seem to be a direct argument for the identity of functional properties with physical properties. Kim presents the functional model of reduction as follows: To recapitulate: to reduce a property M to a domain of base properties, we must first “prime” M for reduction by construing, or reconstruing, it relationally or extrinsically. This turns Minto a relational/extrinsic property. For functional reduction, we construe Mas a second-order property defined 4 Kim quotes this passage as well. The Canberra Plan neglects ground by its causal role- that is, by a causal specification H describing its (typical) causes and effects. So Mis now the property of having a property with such and such causal potentials, and it turns out that property P is exactly the property that fits the causal specification. And this grounds the identification of M with P. M is the property of having some property that meets specification H, and P is the property that meets H. So M is the property of having p. But in general the property of having property Q = property Q (pp. 98-9) · To say that M is the property of having some properry that meets specification His to say that Mis a second-order property in the sense used here. A first-order properry in the sense used here is one that does not have a true characterization in terms of having some other properties that meet a certain sort of condition. So, obviously, a second-order property cannot be identical to a first-order property. A functional property can be thought of as a special case of second-order property that is constituted by the having of some other properties that have certain causal relations to one another and to inputs and outputs.’ The “other” properties are known as the realizers of the functional properties. When Kim says that P is the property that meets the specification H, he is saying that Pis the realizer of H, and of course realizers can be (and are assumed here to be) first order.6 So, it may seem that Kim is arguing for a straightforward contradiction: that a second-order property is identical to its first-order realizer. Later, I describe more of the context surrounding Kim’s argument, which reveals that the natural interpretation just given is not the right one. The passage is misleading, as I will explain. I have been describing armchair philosophical views that purport to {be versions of reductive physicalism. However, there are more sciencebased versions of these views that seem to suggest grounding the mind in functional or computational properties. Recent neuroscience is strongly ,:,:’computational, and a computational view of the mind is often seen as a version of a functionalist view of the mind. So, a view of the mind in neu.J”Oscientific terms would seem to be both physicalistic and functionalistic. Further, it is often said that all science is functional. Lewis (1970) held that all terms of science should be defined functionally. Daniel Dennett (2001) ..,says that functionalism is true generally, for all of science, and that the \most general functional descriptions are at the level of physics: 5 See Kim (1998b, p. 20). I have sometimes defined a functional property as a property that is functional in the sense in the text and that in addition involves causal relations to inputs and outputs. Of course there is no issue of fact as between these definitions. A suggestion to the contrary is made in Block (1990), —————————-·-·-The Canberra Plan neglects ground III no NED BLOCK Functionalism is the idea enshrined in the old proverb: handsome is as handsome does. Matter matters only because of what matter can do. Functionalism in this broadest sense is so ubiquitous in science that it is tantamount to a reigning presumption of all of science. (p. 233) Physics is of course a science and also physicalist, at least about the entities with which it is concerned, so according to the point of view just mentioned, physics is both physicalist and functionalist. The mental includes – at least – states, events, processes, entities, and properties. In the way oflooking at the mind-body problem that I will be promoting, properties are key, and so I focus on them. Why are properties important? The main reason is that from a physicalist perspective, phenomenal similarities must be grounded in physical similarities. Similarities are just shared properties, so of course properties are important. Also, properties are the locus of an important issue concerning causation. The Queen of the Night sings “Der Holle Rache kocht in meinem Herzen,” which shatters the glass. But her words cause the glass to shatter in virtue of their volume and frequency rather than in virtue of their semantic properties (Dretske, 1989; Sosa, 1984). More generally, when one event causes another, some properties of the cause are causally efficacious in respect to (certain properties of) the effect, and others are not. Since functionalism is a causal thesis, there is good reason for a discussion of the functionalist approach to the mind-body problem to pay attention to properties. For simplicity, I take state types to be properties, albeit temporarily instantiated properties, and I think of an event (which I won’t discuss much) as a thing’s having a property at a time. III. Brief refresher course This section provides some elementary exposition on what functionalism is. Readers who are familiar with this material might still read the last two paragraphs of the section.7 Suppose we have a theory of mental states that specifies all the causal relations among the mental states, sensory inputs, and behavioral outputs. Focusing on pain as a sample mental state, it might say among other things that sitting on a tack causes pain and that pain causes anxiety and the pain 7 Readers who wish to see a longer exposition could look at Block (1997b). A somewhat revised version is available at http://www.nyu.edu/gsas/dept/philo/faculty/block/papers/funcrionalism.pdf and in volume 1 of my collected papers (Block, 2007a). jointly with the anxiety cause saying “ouch.” Let us agree for the sake of the example to go along with this silly theory. Functionalism would then say that we could define pain as follows: being in pain = being in some state, which is caused by sitting on tacks, and which in turn causes some other state, and the two states jointly cause “ouch.” The two “so mes” indicate existential quantification, which makes the definition second order, not in the sense of a property of properties, bur in the sense of a property that has a true definition in terms of having some other properties that meet a certain condition. Making the quantification over states more explicit: Being in pain = Being an x such that x is in pain = Being an x such that 3P3 Q (being stuck by a tack causes P & P causes Q and P and Qjointly cause emitting “ouch” & xis in P). 8 More generally, if Tis a psychological theory with n mental terms of which -the 17th is “pain,” we can define “pain” relative to T as follows – the F1 • •• Fn are variables that replace then mental terms; and i1 , etc., are the input terms (such as “being stuck by a tack”); and o,, etc., are the outpur terms (such as “emitting ‘ouch”‘): Being in pain = Being an x such that x is in pain = Being an x such that 3F, … 3Fn [T(F, . .. Fn, i,, etc., o,, etc.) & x is in F,7 } In this way, functionalism characterizes the mental in nonmental terms, that is, in terms that involve quantification over realizations of mental states but no explicit mention of them; thus functionalism characterizes the mental in terms of structures that are racked down to reality only at the inputs and outputs. In characterizing the mental in nonmental ;terms, functionalism gains what has been seen as a benefit of behaviorism while nonetheless acknowledging mental states by quantifying over their ‘realizations, and thereby improving on behaviorism. It is often easier to think about the relation between first- and secondrder properties by using an example of a simple disposition, for example, ormitivity. Dormitivity can be construed as a second-order property, the roperty constituted by the having of some first-order property or other hat causes sleep. Of course, one could equally well construe dormitivity a first-order property, the property of just causing sleep. But now that I ‘,ave acknowledged that there is a first-order construal of dormitivity, the ‘.fhe symbol 3 stands for ~there is.” 3xFx means there is something that is F. 3P3 Q [sitting on a ~causes P & P causes Q} can be read as: there are two properties such that sitting on a tack causes o:he of them and it causes the other. II2 NED BLOCK reader may wonder what the difference is and why anyone would construe dormitivity as a second-order property. In the first-order construal, the property F is dormitive just in case F causes sleep. But if we want to ascribe dormitivity to dormitive things, for example, pills, we have to use the second-order sense. What it is for a pill to be dormitive is for it, the pill, to have such an F that causes sleep, that is, what it is for the pill to be dormitive is for it to have some property or other that causes sleep. That is, x is a dormitive pill if and only if 3 G ( G causes sleep & x has G) -or putting chis so as to eliminate the free variable, dormitiviry in the sense in which it applies to pills is the property of being an x such that 3 G ( G causes sleep & x has G). That is, dormitivity = (h)(3 G [ G causes sleep & x has G]). (We could also think of a pill as dormitive just in case it, the pill, causes sleep, but recall that I mentioned at the outset that I would focus on properties both for metaphysical purposes and because properties are important in causation, so this construal is not relevant to the purpose at hand.) The homes of the two construals are in application to different types of items. This point applies straightforwardly to the functionalist perspective on mentality. If we want a functional definition of mental property terms that apply to properties, the first-order variant will do. For example, the pain-property can be thought of as the property of jointly causing certain outputs together with certain other (mental) properties, being caused by certain inputs, and so forth. But if we want to ascribe those properties to people, we need second-order properties. What it is for a person to have pain, according to the functionalist, is for the person to have some property or other that has certain causal relations to other (mental) properties and to inputs and outputs. IV. Can a second-order property he a first-order property< A second-order property is one that has a true characterization in terms of having some other properties that meet a certain condition. A first-order property is one that does not have such a true characterization. So, it is just a contradiction to claim that a second-order property is a physical property. Why do some appear to think otherwise? Dormitivity -construed as a second-order property -has first-order chemical-realizing properties such as (having the) structure C”H”N,03 (phenobarbital) that causes sleep. What is the relation between a secondorder property and the disjunction of its first-order realizers if not identity? By the disjunction I mean the property that consists in being C”H”N’ 03 (phenobarbital) or in being C,6H,3ClN, 0 (diazepam), or … But what The Canberra Plan neglects ground n3 does the ” … ” mean? It seems that the ” … ” means something like: “or some other first-order properties that cause sleep.” But the “some” reveals that the supposed first-order disjunctive property is really second order. Note that the identity claim amounts to something like this: the property constituted by being an x such that 3F (F causes sleep & x has F) = the property constituted by being an x such that (x has P and P causes sleep) or (x has Q and Q causes sleep) or x has some other property (maybe more than one) F such that x has F and F causes sleep. Again, the “some” shows us that the property expressed is second order. But perhaps we can do without that clause with the “some” in it? We can just list the disjuncts. Suppose that there are exactly two first-order dormitive structures as a matter of physical law, C,,H,,N,03 and C,6H,3ClN, 0. However, if there were another first-order property that caused sleep, it would be dormitive, according to the second-order definition, without being one of the first-order disjuncts. And a similar point holds for the · infinite disjunction. So, there is a modal difference. Even withour a modal difference, the hyperintensionality of grounding ‘leads to a similar conclusion. The existence of Socrates is the ground of the existence of the singleton of Socrates rather than vice versa (Fine, 2012) ‘4espite these facts obtaining in exactly the same worlds. The application {is this: even if somehow a first-order physical property played the pain nctional role in every possible world and nothing else played that role in y possible world, there would still be a question of whether the obtaining fthe physical property was the ground of the obtaining of the mental !operty. That is, even if the mental and physical properties are coextensive cross all possible worlds, the question still arises as to whether the mental toperty is instantiated in virtue of the instantiation of the physical roperty. I give an example toward the end of the chapter -in which the hysical property is indexical -that should make this point vivid. The reader may feel that the fact that a second-order property cannot be entical to a first-order property can mislead us with regard to physicalism, f the second-order property can itself be seen as in effect first-order ysical so long as all its realizers are first-order physical. (In addition, it uld have to be stipulated that there are no extra nonphysical “ghost” ‘ental properties.) I explain the inadequacy of this view in the next section. V. Metaphysics, ontology, and disjunctive ground e the term metaphysics to mean the study of ground, and I use the term l.ogy to concern what types of things exist. My use of the term ontology ·ves from Quine (1948). Quine and subsequent discussions influenced 114 NED BLOCK by him speak of a person’s or of a theory’s ontological commitments, meaning commitments on what types of things exist. Sample ontological disagreements concern whether universals or souls or absolute space exists. The ontological issue of what it is like to experience pain is whether in adopting an ontological commirment to the experience of pain we adopt an ontological commirment to anything immaterial. An ontological physicalist says no. A metaphysical physicalist, by contrast, claims that the ground of the experience of pain – what all experiences of pains have in common in virtue of which they are experiences of pains – is a physical property that explains the experiential commonality of experiences of pains. Metaphysical physicalism could fail even if ontological physicalism is true. The phenomenal commonalities berween different pain-feeling creatures could fail to have a physical ground even without any immaterial souls. Metaphysical physicalism could be true even if ontological physicalism is false if, for example, our material minds have an immaterial adjunct that is part of a communication nerwork with angels rather rhan part of rhe ground of our mental properties. Dualism and physicalism are naturally understood as both ontological and metaphysical theses, but functionalism can be a metaphysical thesis without being an ontological thesis. Let me explain. Dualism and physicalism disagree on whether there is anything immaterial. But functionalism is compatible with both ontological positions because functionalism takes no stand on the occupant of the functional roles that define it. Functionalism can say that the ground of pain, whar makes rwo pains both pains is a common functional role. This is a metaphysical, not an ontological, doctrine. Pains could have that functional role whether or not they involve nonphysical substances or properties, so long as the nonphysical substances or properties are causally efficacious in regard to other states, inputs, and outputs in the right ways and can be causally affected in the right ways. Suppose there are souls in some adding machines that make them work. Still, the ground of something being an adding machine – to the extent that one can speak of something so nominal as a ground at all – is that adding machines’ states function so as to add. If the soul sruff can function in this way, it doesn’t make the metaphysical narure of adding in any way nonfunctional. There is nothing about the function that constirutes adding that requires a material basis. Similarly, a metaphysical functionalist should say that the existence of souls (ontological dualism) need not be relevant to the metaphysical issue of what grounds pain – the answer could be functional just as with adding. The Canberra Plan neglects ground n5 As I mentioned, the physicalism of David Lewis (1966, 1970, 1972) derives from the idea that pain can be defined, a priori, on the basis of its causal role, R. Brain state B in us has R as a matter of fact, so pain = B. Suppose further that there are no nonphysical pains and that there never have been and never will be any nonphysical pains (Lewis, 1994). The result is a kind of ontological physicalism, but note that it does not amount to a metaphysical physicalism that grounds mentality in the physical. What is common to Martian pains, if there are any, and octopus and human pains that grounds the fact that they are both pains is not anything physical, on Lewis’s perspective, but rather the fact that they are all instantiations of causal role R. So, on the metaphysical question of what grounds mind, Lewis (1966, 1970, 1972) should be seen as having no view, or perhaps as being a functionalist rather than a physicalist. In his preferred regimentation (not including Lewis, 1980, in which he adopts the weird “mixed” theory that I will be getting to), pain is physical, but having pain is identical to a second-order (functional) property, namely, the property of having some state or other that plays causal role R. A context-relative definire description of the form “the state that has causal role It picks out one physical state in us, another in Martians, and so on. But “having pain” (on Lewis’s regimentation) is not a context-relative designator bur rather a rigid designator that always picks out the same second-order property, namely, the property of having some realizer or other that satisfies causal role R. The upshot is that Lewis is an ontological physicalist and, to the extent that he had any metaphysics of mind at all, a metaphysical functionalist. Note, incidentally, a point emphasized by Kim (1972), that the physical basis of pain can be sufficiently abstract so as to be shared by humans and octopi just as rwo physically very different substances can have the same temperatur