A critique of Jaegwon Kim.doc

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 1/100

A critique of Jaegwon Kim's meditations on the mind/body problem

(thing) by frankdeluxe (2.1 hr) (print)   ?  2 C!s Mon Sep 27 2004 at 17:56:28

1. Opening Shots 

In  Mind in a Physical World , Jaegwon Kim wrestles with the problem of 

 positing mental life and mental events as causally efficacious in a world thatis fundamentally ontologically physical. By "ontologically physical", I mean

that we generally tend to evaluate events in the world according to the laws

of physics as they have been proposed and revised over years of scientificinquiry. In this kind of a world (in our world), talking about "mind" as

something independent of material substance and physical laws is

troublesome. Since we can usually associate some kind of neural or  physiological state with any given mental state, we are inclined to simply

refer to the former rather than the latter when engaging in causal

explanation. In other words, brain states are causally sufficient to explain

 physical events, while mental states are not - especially when their corresponding brain states get the job done just as well. Of course, this

leaves us with the problem of whether or not the mental life and its events

are illusory or epiphenomenal. Kim's project is an attempt to preservementality, with its intentional dispositions, so that we are not forced to

simply reduce all intentionality, autonomy, and agency to a byproduct or 

misinterpretation of  neurophysiological events and states. By the end of 

 Mind in a Physical World , he is forced to concede that, given the assumptions about physicalism upon which he is working, there is no

real way to fully preserve the mental life. On his view, we can go with

either a revival of property dualism  (which will also revive its own set of  problems) or  some variety of reductionism (which essentially denies the

reality of mental life).

I would like to suggest that despite Kim's efforts at developing a new kind of causal talk that includes mentality, he remains trapped in a framework that

will never allow it. This framework owes much to -and is, in fact, dependent

upon- Cartesian substance dualism, which is precisely what it tries to leave

 behind. My aim in this paper is to discuss the problems of Kim'sontology, and to try and present an alternative. In books like Mind and 

World and Mind, Value, and Reality, John McDowell presents an ontological

and epistemological account in which the distinctions between the internaland external (for our purposes, the mental and the physical) are broken down

(or at least softened). For McDowell, there can be no stark distinction

 between the mental  and the physical , and neither one can be reduced to theother's terms. It is this kind of ontology that I would like to propose that we

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 2/100

look at adopting before we start talking about mental events as causal

events. I will refer to McDowell and throughout this paper, as well as to

Charles Taylor and Hilary Putnam, who write in support of him. Also, I willrefer to  Donald Davidson, with whom Kim is an interlocutor (but still

something of a kindred spirit) in this debate.

2. Ontology: Descartes and his footnotes 

Descartes' dual substance ontology holds that there are two types of  properties: the material (or physical) and the mental. They are both present

in human beings and are separate, yet related. He writes that although

 perhaps ... I have a body that is very closely joined to me, nevertheless,

 because on the one hand I have a clear and distinct idea of myself, insofar asI am merely a thinking thing and not an extended thing, and because on the

other hand I have a distinct idea of a body, insofar as it is merely an

extended thing and not a thinking thing, it is certain that I am really distinctfrom my body, and can exist without it. {MFP 51}

For Descartes, the mind and the brain are independent of one another but

they interact causally; the brain perceives sensations from the rest of the

 body and transmits them to the mind. The mind, in turn, interprets the

information and directs the body through the brain {Ibid. 56-7}. The

 powers of imagination, understanding, willing, and so on are specifically

 powers and properties of the mind; Descartes does not believe that they are brain functions. The autonomy of the subject is assured by the mind's

separation from and sovereignty over the brain and the rest of the body.

Unfortunately, the biggest problem with Descartes' account is that he cannotaccount for the nature of the relation between mind and body. This puts

the mind's causal role into serious doubt. Moreover, Descartes further 

complicates matters by stating that there is a law-like or nomological

relationship between mental and physical events: "since any given motionoccurring the part of that part of the brain immediately affecting  the mind

 produces but one sensation in it, I can think of no better arrangement than

that it produces the one sensation that, of all the ones it is able to produce, ismost especially and most often conducive to the maintenance of a healthy

man" {Ibid. 57}.

Descartes is proposing a very specific relationship between physical events

(or kinds of physical events) and mental events (or kinds of mental events).Humans are constituted such that sensations will always cause the same 

kinds of mental events - this is a law-like, or nomological, relationship . As

we have seen, Descartes' inability to account for the nature of thisrelationship has left his ontology, metaphysics, and approach to mind-body

causation vulnerable to arguments that bracket out the mental in favour of  

the physical: "His problem, as his contemporaries saw, was to show how hisall-too-common-sensical thesis of mind-body interaction was tenable within

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 3/100

his ontology of two radically diverse domains of substances, minds, and

bodies" { MPW 29}. In other words, his whole account can be resolved

by mind-body theories that reduce mental states to their corresponding

neurophysiological states. This, of course, is reductionism: "to reduce a

 property, or phenomenon, we first construe it-or reconstrue it-functionally,in terms of its casual/nomic relations to other properties and phenomena"{Ibid. 25}.

The move to physicalism basically consists of rejecting the ontological

 plausibility of a non-material substance such as mind. Indeed, it seems

almost inevitable, given the movement away from transcendental andspeculative philosophy in the last 400 years or so. As Charles Taylor writes

in the essay "Foundationalism and the Inner-Outer Distinction", maintaining

a firm belief in an autonomous and independent mental life seemsimpossible "if we conceive of spontaneity as a kind of limitless freedom,

which at the point of contract has to hit a world under adamantine, post-Galilean `laws of nature'" {FD 109}. By spontaneity, he of course refers tothe kind of causal efficacy and power that a mind holds when its eventsand states are not reducible to physical events and states. In our "post-

Galilean" world (that is, a world which operates largely on the assumptions

and schemes provided by the scientific method), the mental substance of 

Descartes' dual substance ontology is crossed out, or at least reduced to

being a function of material/physical properties. The problem is one of 

making room for a robust and complete mental life in a world that doesn'tneed it and cannot ontologically and conceptually sustain it.

There have been numerous attempts to make room for mentality: the main

thrust of philosophers like Kim seems to be dealing with both the problems

of exclusion and of reduction. The exclusion prob lem arises for physicalistthinking when we argue, "mental event m, occurring at time t , causes

 physical event p, and ... this causal relation holds in virtue of the fact that m

is an event of mental kind  M and  p is an event of physical kind P" {MPW37}. Since we look for and require physical explanations in our physical

world, we are bound to assign some p that is nomically linked to any given

m (Descartes already implied as much in the Sixth Meditation). It does not

seem to be physicalism itself that sounds the death knell for mentality;rather, it is the necessity of reductionism,  which arises because we require a

nomic relation between the mental and the physical. This is what Kim refers

to as "type physicalism", which "reductively identifies mental propertieswith physical properties, implies mind-body supervenience" {Ibid. 12}.

Mind-body supervenience, of course, is the theory that deals with "a

relation between two sets of properties, the supervenient properties and their 

 base properties" {Ibid. 9}. Kim focuses on what he calls strongsupervenience {Ibid.}, and applies it to the mind-body relationship thusly:

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 4/100

Mental properties supervene on physical properties, inthat necessarily, for mental property M , if anything hasM at time t , there exists a physical base (or subvenient)property P  such that it has P  at t , and necessarily 

anything that has P at a time has M at that time. {Ibid.}

Here we can see the problem: since  M  has and must have  P  as its base

 property, we can construe causal talk such that P is also causally efficacious

in terms of bring about physical events in the world. We can also

construe it such that P is a full cause of a physical event, which makes it

clash with M : if our desire for water (to use Kim's example in Chapter 3) can be explained just as well by a neurophysiological account as it can by an

account involving intentional states, desire, and so on, then the

neurophysiological account can suffice as causal explanation . In this case,the mental is excluded; certainly, it makes no sense to exclude the physical

explanation, for without the physical, we cannot have the mental, either.This type of relation, of course, is also dependent upon the type-type (that is,

types of mental properties are supervenient upon types of physical properties) nomic relationship involved in "strong supervenience"; there are

other possible supervenience theories available.

Kim notes that Donald Davidson takes another approach, one that does not

appeal to strict nomic relations between sets of properties. In the essay"Mental Events", Davidson rejects the idea that there are strict laws between

mental and physical events and instead takes up the view he calls

"anomalous monism". On this view, mental events are not reducible to physical ones; it "rejects the thesis, usually considered essential to

materialism, that mental phenomena can be given purely physical

explanations" {ME 214}. Further, he says that anomalous monism allowsfor a different kind of supervenience:

Such supervenience might be taken to mean that therecannot be two events alike in all physical respects butdiffering in some mental respect, or that an objectcannot alter in some mental respect without altering insome physical respect. Dependence or supervenienceof this kind does not entail reducibility through law or definition: if it did, we could not reduce moral propertiesto descriptive, and this there is good reason to believecannot be done; and we might be able to reduce truth ina formal system to syntactical properties, and this weknow cannot in general be done. {Ibid.}

In Davidson, at least, we have some possibility of adopting asupervenience theory  that tries to preserve physicalism as well as theirreducibility of the mental. In a similar spirit, Kim proposes the

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 5/100

multiple realization thesis. Multiple realization, like Davidson'sanomalous monism, rejects the notion that there is a strict nomicrelation (or, as Kim puts it in Chapter 4, "a biconditional bridge law of the form P ←→Q" {MPW 92}, which means that P and Q must always

occur together) between mental properties and their base physicalproperties. In other words, there can be a whole array of physicalevents and states potentially at work behind any given mental state.In this case, M  is not just nomically linked to P ; it can be realizedthrough properties {P 1 ... P n}. On the multiple realization view, we canavoid reducing mental states to physical ones in the way implied bystrong supervenience.

3. Kim's futile struggle 

Unfortunately, the multiple realization argument does not prevent the

exclusion problem, and Descartes indeed has his revenge. Even thoughmultiple realization complicates the supervenience picture, there comes a

 point at which we have to admit that any possible P that acts as a realizer for 

 M  is also in the position of excluding it  {Ibid. 110-11}. As I stated at theoutset of this paper, by the time he reaches the end of the book, he has to

throw in the towel and accept some form of reductive physicalism:

What is becoming increasingly clear from the continuingdebate over  the mind-body problem is that currentlymiddle-of-the-road positions , like property dualism,anomalous monism, and nonreductive physicalism, arenot easily tolerated by robust physicalism. To think thatone can be a serious physicalist and at the same timeenjoy the company of things and phenomena that arenonphysical, I believe, is an idle dream. Reductivephysicalism saves the mental but only as a part of the physical. If what I have argued in these lectures isin the right ballpark, that is what we should expect fromphysicalism. And that is what we should have expectedall along. Physicalism cannot be had on the cheap.{Ibid. 120} me parece es kim

He examines various forms of reduction: functionalism, eliminativism,

epiphenomenalism, and so on; however, they all undermine the mental infavour of the physical. There is no way to really preserve the mental,

given the commitment to physicalism with which Kim is working. Even

though he says in Chapter 2 that we need to make a choice between

"various metaphysical alternatives, not between some recondite

metaphysical principle on the one hand and some cherished epistemological

 practice or principle on the other" {Ibid. 62}, Kim is falling prey to the samekind of trap ontologically. What I mean is that while he is trying to evaluate

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 6/100

different possible metaphysics (different accounts of how things

fundamentally  work  (funciona)), he is not thinking about different

possible ontologies (different accounts of what things fundamentally are).The physicalist ontology is a descendent of Descartes, despite its attempts to

escape from the problems of Cartesian substance dualism. While Descartes'ontology affirms and posits two radically different substances, physicalism

only posits the existence of one, while still maintaining the opposition . In

other words, the difference seems to be that Cartesian ontology accepts the

intuitive notion that there is a mental life above and beyond physical

 properties but cannot account for the relation, whereas physicalist

(especially reductive physicalist) ontology tries to only affirm the

physical but cannot keep the mental from creeping back in and causing

problems.

It does not really make sense to doubt the existence of the mental

outside of philosophical inquiry; Kim says as much in Chapter 3, when heagrees with Tyler Burge and Lynn Redden Baker and states that the issue is

not "one of choosing between metaphysics and mental causation: most of ushave already chosen mental causation" {Ibid. 62}. So, he is quite right in

coming to the conclusion that his philosophical exercise in  Mind in a

 Physical World  has not produced a satisfactory account of  mind-body

causation; however, he cannot seem to see a way out of it, other than going

 back to trying some form of dualism out {Ibid. 120}. This is where Kim's

account is most unsatisfactory, since he does not explore other

possibilities of construing the world; however, given that his project in the

 book was to examine the role of the mind in physicalist ontology, it is not as

though he was obliged to do so. I would suggest that he merely fails by not

suggesting anything outside of physicalism or dualism,  when there are

other options available. Here I would like to turn to the work of  John 

McDowell and the next section of the paper, in which I try to constitute

some kind of alternative proposal that may allow us to speak about the

mind as causally efficacious once more.

4. McDowell and blurring the boundaries 

The problem created by Descartes is, of course, the radical opposition

 between mental and physical, subject and object, internal and external. Justas we as we subjects are opposed to the external world as object, our minds

are the true seat of our subjectivity and our bodies play the role of the object.

As Descartes says, it is the mind that is the true self: the body is secondary.

This is at the root of the philosophical tradition within which mainstream philosophy of mind (not to mention metaphysics and ontology) takes place.

It carries with it not only the problem of mind-body interaction but also the

 problem of  how the world is constituted. This, of course brings up thequestion of how we are constituted in the world, whether there actually is

a subject/object or internal/external opposition. The notion that the mind is

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 7/100

somehow the self  in a truer way than the body seems to be a bit of a

 problem. Certainly John McDowell  phrases it in this way:

If we begin with a free-standing (independent) notion of 

an experiential route through objective reality, atemporally extended point of view that might bebodiless so far as the connection between subjectivityand objectivity goes, there seems no prospect of building up from  there to a notion of a substantialpresence in the world. If something starts outconceiving itself as a merely formal referent for  "I" (which is already a peculiar notion), how could it cometo appropriate a body, so that it might identify itself witha particular living thing? Perhaps we can pretend tomake sense of the idea that such a subject might

register a special role played by a particular body indetermining the course of its experience. But thatwould not provide for it to conceive itself, the subject of its experience, as a bodily element in objective reality-as a bodily presence in the world. {MW 103}

McDowell wants to say that positing a "true" nonphysical self is

nonsensical: we require bodily instantiation in the world for experience,which not only makes mentality possible but also provides us with the

content to develop a sense of "self" and a relation to objective reality. For 

McDowell, there is no rigid opposition between the mental and physical

self . As he says in the afterword to Mind and World , "there is no obstacle to

supposing that my use of `I' as subject refers to the human being that I am"

{Ibid. 179}. To reverse this proposition, we may ask the question of whether or not it actually makes sense to posit the  physical self (that is, what we can

observe in terms of neurophysiological states, for example) as being more

"real" or "true" than the mental self  (that is, our intentional states, beliefs, desires, and accumulated experience). We might ask the question of 

what doubting our own mentality becomes when even our doubt is reducible

to a neural firing: if our doubt is nothing more than an epiphenomenon of 

some physical event, then how can we even say that we are doubtingwhether or not we have a mind? In that case, what are we even talking

about? Why is it necessary to make such a strong distinction between what

we observe of our  neurophysiological states and the content of our

intentional states?

The point is, as Taylor says (in support of McDowell), " The I/O image is

too powerfully embedded in our beliefs and (scientific, technical, freedom-

oriented) way of life" {FD 107}. We are used to conceiving the worldthrough a theoretical point of view: we evaluate it according to scientific

inquiry and reach our conclusions for better or for worse. Since mentality

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 8/100

(as mentality, not as a function of neurophysiological states) eludes the 

explanatory framework provided by the laws of physics, it is not accorded

causal efficacy or relevance within that framework. It is treated as an

incoherent notion and thus relegated to the realm of epiphenomena. What

we are not used to doing (in this branch of and approach to philosophy, atleast) is taking our mental states and events for what they are. Because of thescientific orientation so deeply embedded in us, it does not seem

philosophically kosher (autorizado) to phenomenologically evaluate our

mental states and avoid recourse to the kinds of problems inherent in

adopting robust physicalism.

McDowell wants to develop an ontology of the world that proceeds

 phenomenologically rather than according to some a priori conception of the

self or according to some detached, theoretical model. This is notnecessarily meant to reject neuroscience; however, he rejects adopting

that sort of model as the metaphysical or ontological schema of theworld.  To be sure, intentionality makes no sense on the scientific view - 

except as it can be described scientifically. This involves describingintentional states in such a way that only what is scientifically observable

can be included.

What McDowell thinks is lost in all of this is that this description does not 

represent a view of the world that has to be more fundamentally true thanother possible descriptions. Reductionism does not make sense: it cannot

account for the content of peoples' lives, and it still presents a "closed" view

of what it is to be human. To clarify, I mean that it reductionist thinking

locates the main locus of mental activity in brain states, and due to its

dependence upon the scientific method, it cannot make sense of 

intentional states. McDowell writes that a "deep element in a broadlyCartesian outlook is an inability to conceive `cogitation' as part of something

`merely' natural (so the cast of thought will incline us to put it), such as the

life of an individual animal. This inability is manifested in the Cartesiansegregation of `cogitation' into a special realm of reality" {MVR 381}.

Conversely, a deep element in a robustly physicalist outlook is an inability

to conceive of `cogitation' as sufficiently natural-since it cannot be described

in any other way than through neurophysiological states; furthermore, the physicalist outlook does not account for the fact that its grasp and dominion

over what is and is not natural is always open for negotiation (to play off a

comment made by Kim in MPW). To say that neurophysiology-or even

 physics   proper-can account for  all that is natural is to do two thing s .

Firstly, it is to conceive of nature as something that can only be observed,

described, and accounted for by science. Secondly, it is to forget that our 

scientific vocabulary and practices are not a priori or ahistorical. We have

not received the laws of physics from some vantage point that is exterior

to our experience of the world.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 9/100

By that token, McDowell notes that modern science understands its subject

matter in a way that threatens, at least, to leave it disenchanted, as Weber  put

the point in an image that has become has become a commonplace. Thisimage marks a contrast between two kinds of intelligibility: the kind that is

sought by (as we call it) natural science, and the kind we find in somethingwhen we place it in relation to other occupants of `the logical space of 

reasons', to repeat a suggestive phrase from Wilfred Sellars.

If we identify nature with what natural science aims tomake comprehensible, we threaten, at least, to empty itof meaning. By way of compensation, so to speak, wesee it as the home of a perhaps inexhaustible supply of intelligibility of the other kind, the kind we find ina phenomenon when we see it as governed bynatural law. It was an achievement of modern thought

when this second kind of intelligibility was clearlymarked off (demarcado)from the first. In a commonmediaeval outlook, what we now see as the subjectmatter of natural science was conceived as filled withmeaning, as if all of nature were a book of lessons for us; and is a mark of intellectual progress that educatedpeople cannot now take that idea seriously, exceptperhaps in some symbolic role. {MW 71}

In this passage, McDowell adeptly (profundamente, completamentecompetente) brings the problem to the fore (adelante, destaca): oneontological conception of the world has overtaken (superar)another , and it has produced the kinds of problems with whichphilosophers like Kim are still struggling today. Hilary Putnam writesin agreement with McDowell, saying that McDowell "speaks of aneed for a naturalism which `re-enchants nature,' (and) he isdeliberately coupling the issues of accepting the intentionalnotions and the normative notions used in connection withnatural science (the epistemic norms, as it were) without reductionand without apology with the issue of accepting our ethical notions ...without reduction and without apology" {MMW 187}. Kim himself saysthat "Metaphysics is the domain where different languages, theories,explanations, and conceptual systems come together  and have their mutual ontological relationships sorted out (aclarar, arreglar) andclarified" {MPW 66}; by attempting to reconstitute ontology in aradically different way than what Descartes bequeathed (legar) to us,McDowell is also attempting to come up (mencionar) with ametaphysics that refuses to accept reduction and holds on firmly toall the things which imbue (imbuen, empapan) our lives withmeaning. I would suggest that it is only if we take on robustphysicalism as a precondition for talking about mental causation

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 10/100

and the mind-body problem that we have to find a problem withmultiple forms of causal explanation and posit a conflict betweenmentality qua intentionality and mentality qua physical reduction.

5. You've gotta know when to hold 'em...

With McDowell's approach, talking about mental causation does not have to

 be so problematic. In a manner recalling the early Heidegger, he brings out

(introduce) the way in which  philosophical inquiry transforms  over time

into bodies of doctrine that cease to elucidate aspects of the world and

start to act as limitations on what we can and cannot conceive - or rather,

the ways in which we can and cannot conceive. In ways that Kim does not,

he effectively goes a good distance toward making metaphysics "work"again. Although Kim does put in a good effort, he does not ask a lot of 

questions that are crucial to the project. Significantly, the topic of what

physicalism (as well as the "natural") should or should not encompass(abarcar); namely, whether or not the physical really only falls within the

 bounds of what can be described by natural science, and whether or not there

is adequate justification (or even a good commonsense  reason) for adoptingthe vocabulary of natural science as our primary ontological category.

Philosophical tinker ing (reparador)(that is, simply combing over and trying

to reconstrue the very concepts that will not allow us to get to where we

want to be) is clearly not enough if we are to save the mind fromreductionism; a much more bold (audaz, llamativo)stance must be taken, and

much more radical changes need to be made in order to make causal talk 

appropriate for our phenomenological and intuitive experience of the

world.

Although this paper by no means can offer a complete program for making

these changes happen, I would suggest that we can see an open doorway not

only through John McDowell's work, but also through the work of writerssuch as Taylor, Putnam, and even Davidson; these thinkers (and I am in

support of their work) recognize the inadequacy of the "philosophical

tradition" and its ontology, and they are trying to work toward an ontology

that works with our mode of existence. That mode of existence is, of 

course, as human beings in the world, whose experiences, beliefs,

intentions, and actions cannot sensibly be reduced to scientific propositions.Such talk not only undermines (socava, aruina, desaleienta) itself (by

reducing its very content to a product of epiphenomena), it is also not

suitable for understanding what it is to be human. Being human necessarily

involves holding beliefs, experiencing real content in the world, and

acting upon intentional states that do indeed "make a difference". We

need to have an ontology that reflects this reality, and the task is underway.

(en movement, en progreso)

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 11/100

Works Cited

• Davidson, D. "Mental Events" {ME}, in Essays on Actions and Events, 1980;

Oxford: Clarendon Press, 2001.•

Descartes, R. Meditations on First Philosophy  {MFP} trans. Donald A.Cress, Indianapolis: Hackett Publishing Company, 1993.• McDowell, J. Mind and World {MW}, Cambridge: Harvard University Press,

1994.• ---------- Mind, Value, and Reality  {MVR}, 1998; Cambridge: Harvard

University Press, 2002.• Putnam, H. "McDowell's mind and McDowell's world" {MMW}, in Reading 

McDowell: On Mind and World , ed. Nicholas H. Smith, New York:Routledge, 2002.

• Taylor, C. "Foundationalism and the inner-outer distinction" {FD}, in

Reading McDowell: On Mind and World , ed. Nicholas H. Smith, New York:Routledge, 2002.

www.everything2.com/index.pl?node_id=1677621 - 54k

-ON THE MATTER OF MINDS AND MENTAL CAUSATION

At a first blush, materialism and explanations involving the mind appear to be

incompatible. For anything that is explained using mental states would have another 

explanation using lower level physical states. The physical explanation would trace theactual causal history, while any mental explanation would be redundant, derivative,

explanatorily superfluous. If we are materialists, then it seems that we should simply do

away with the mind, as it were. (See, e.g., Broad 1925; Campbell 1970; Feigl 1970;Goldman 1969, Honderich 1988; Kim 1979, 1989, 1990; Macdonald and Macdonald 1986;

Malcolm 1982; Shiffer 1989; Sosa 1984).

But on the other hand, mental events appear to be separate from physical ones. Multiple

realizability -- the fact that any particular mental state could be instantiated by an indefinitenumber of lower level physical configurations -- tells us that no lower level particular 

 physical state is necessary for any mental one (Fodor 1980, Putnam 1967; see also

Davidson 1993). This precludes any interesting type-type reduction between the mind and

the brain (Fodor 1974, Putnam 1976; see also Wimsatt 1976). But without this sort of reduction, it is difficult to see how any purely physical explanation could explain mental

 phenomena.

This of course is just a version of the mind-body problem with which dutiful materialistscontinually struggle. Our metaphysics breeds epiphenomenalism, but our mentalism breeds

a different ontology. Hence, the epistemological difficulties. . . .

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 12/100

On a second pass though, the story is a bit more complicated. If we want to be materialists

first and foremost, then we can say that some mental state caused some physical event just by noting that each mental token is identical to some physical token or other and that the

 physical world is causally closed. However, what a materialist cannot maintain is that it is

the mental state of a given type qua mental state of that type that causes a physical change.

(At least, a materialist cannot maintain this without additional labor, which I try to provide below.) That is, a materialist seems compelled to assert that every mental property  M has

some physical implementation  P such that we can subsume occurance of  P under some

 physical law. To pretend otherwise would be to materialism. But, as van Gulick writes,"once we acknowledge that there is such a . . . physical law . . ., like a young cuckoo in the

nest of a robin, it 'greedly'; grabs off the causal potency for itself" (1993, p. 241). The fact

that  M  seems to cause some property  N  so that  M  -- >  N is lawlike results from our more basic physical causal laws and the fact that  M   supervenes on the physical. LePore and

Loewer conclude:

The real locus of causal powers are the physical properties . . . . [ M ], so to speak, gets

carried piggyback on physical properties and it is mere appearance that possessing . . . [ M ]determines c';s causal powers. The basic physical properties and laws determine both the

causal relations among events and the non-basic [lawlike relations] . . . . It is merely an

appearance that the non-basic [relations] . . . determine causal relations among events.

(1989, p. 187)

But what materialists seem forced into concluding regarding the causal efficacy of 

mental properties qua mental appears to be fundamentally mistaken. There is an

undeniable difference between someone breaking a glass by accidentally brushing upagainst it and smashing a glass in a fit of anger. In the first case, the person';s cognitive

state has little or nothing to do with the glass shattering,  but in the second case, the mental

state qua  anger  seems to be quite relevant to the broken glass (cf ., Achenstein 1977,

Anscombe 1975, Dretske 1989, Honderich 1983, Searle 1983, Sosa 1984). Of course, thehistories of the two cases diverge once we travel even a bit up the causal chain, so we

would never confuse one case for the other. However, the point here is that in one instance

there is no need to invoke any mental properties in explaining the event, but mental ishighly relevant in accounting for the other. How are we to understand this difference

 between the two cases? What is the proper way to understand the metaphysical and the

explanatory relations among the mind, the brain, and the resultant behavior?

This paper is devoted to exploring these questions. We need to articulate a version of mental causation that rests comfortably with materialism. Here I explore the more popular 

middle ground in which mental phenomena are claimed to be "as real as" other higher level,

causally efficacious, properties. I argue that this position is inappropriate because fails toanswer our epistemological difficulty: How are we do construct proper explanations using

apparently superfluous mental phenomena?

Here is how I proceed. Section I sketches a multi-layered view of the world that allows a

 place for causally efficacious mental events. However, my weak criterion for being a causedoes not help us answer what I take to be the real issue: how to get (and whether we can

get) mental phenomena in proper explanations of intentional behavior. These questions

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 13/100

are highlighted in the next section in which I suggest that putative explanations relying

on the mental are no different from explanations which focus solely on lower level

physical properties. Here I adopt a model-theoretic conception of science. Sections I and

II serve to move the metaphysical question of  whether mental states are causally

efficacious to an epistemological consideration of whether causal chains involving the

mind are appropriate items to include in our accounts of behavior. The final section,Section III, focuses on this question.

I

With the assumption of materialism, we must hold that mental properties just are higher 

order physical properties. That is, mental properties exist at the same level of organizationas some physical properties -- these properties might be fairly complex or abstract, but they

are physical nonetheless. Of course, the lower level physical structures may still be

 primary since they appear to be ontologically prior. Hence, we say that mental properties

depend upon physical ones (and not vice versa) (though see van Gulick 1993 for a contrary

 position). Materialism gives us something like a weak form of property "dualism" withthe lower level physical properties determining the higher level mental ones. We can

understand lower level P ';s as specifying a microstructural property of c that provides

a mechanism for the implementation of  M in c. (See also Fodor 1989, Jackson and Pettit

1990, Kim 1993b, McLaughlin 1989.)

If we assume that both P and M have law-like links to N , and that P is a cause; should we

say that M has a distinct causal role as well? Jaegwon Kim argues that we should lean on

the ontological priority of the lower level physical properties and recognize that the only

way to get mental properties or other higher level social facts is through a lower level

 physiological property (1983; 1989a, b; 1992, 1993a, b; see also Block 1990, Field 1980,Fodor 1981, LePore and Loewer 1989, Segal and Sober 1991, Smart 1963). He denies that

mental properties are proper causal agents in explanation. Simplicity would favor  P alone.

We accept that if  N occurs by being realized by Q, then any cause of Q must be a cause of  N . We can see this, for example, when we treat pain by interfering with the body. Hence , if 

 M is realized by P , then the causal powers of  M are identical to (a subset of) the causal

powers of  P .  Kim concludes that there is no such thing as actual (explanatory) mentalcausation. This effectively abolishes M .

But there are difficulties with this sort of eliminativist picture. If we were to adopt Kim's

view, then we would have to acquiesce that physics is the only science and that all we can

legitimately discuss are vectors in Hilbert space, for it makes all  distinct higher level properties causally inefficacious. It entails that the only causally relevant properties would

 be those defined in physics, and so all chemical, biological, and sociological properties

must be inert. For example, we would not be able to explain the death of AIDS patients

being caused by a decrease of T-cells in their blood, for the biological property "death," thesocio-biological property "AIDS patient," the biochemical entity, "T-cell," and so on, are

not the proper sorts of properties or objects that could engage in causal interactions.

(See Fodor 1989, Humphries 1992, and van Gulick 1993 for different arguments to thesame conclusion.)

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 14/100

I recognize that Kim would not be troubled by my response, since he thinks that there can

 be legitimate explanations using properties that are not causally efficacious but only"causally relevant" (to use Philip Petit's phrase). However, my mind boggles here. How can

something be causally relevant  without also being effiicacious with respect to

something? Perhaps this is a failure of imagination on my part though. Suffice to say for 

now that Kim's position is controversial and, more important, less than optimal, given our goal of articulating a version of honest-to-goodness (real, genuine) mental causation. At the

least, we should strive to do better before admitting defeat.

My diagnosis of Kim's proposal is that he does not take seriously enough the different

levels of organization in the world. One and the same object -- the eye, for example --can be and in fact is described differently depending upon what sort of questions are being

asked. We can talk about the eye in terms of its cognitive function, in terms of its anatomy,

in terms of its physiology, in terms of its chemistry, and so on. Moreover, within each of those broader divisions we find a further hierarchy of descriptions. When discussing an

eye's physiology, we can distinguish the rod from the cone cells, trace their two general

 paths to the retinal ganglion, and point out how the different response patterns of the rodand cone pathways affect what happens beyond. Or we can distinguish among the various

types of rod cells and how they interact with one another in the retinal ganglion and beyond

in terms of the response patterns of particular rod cells. Or we can talk about the firing properties of a particular rod in terms of the influx and efflux of ions. And so on, with each

"level" of description being legitimate with respect to answering certain questions

about eye physiology, but not others.

The picture of the world that emerges with the perspective of different levels of description and organization is obviously a layered one. We can understand this world in

terms of  sets of entities which constitute the domain of particulars for  each level of 

organization. Sets of properties would then be defined over each domain or over sets of 

domains. In contrast to Kim's view, this perspective ignores the ontological priority of thelower level instantiation assumed above.

It is instructive to think of this egalitarian relationship between the higher level  P 's

(including  M 's) and the lower level  P 's as near chaotic phenomena. The sort of larger scale order which emerges in the scroll waves of concentration in the Zhabotinske reaction

(Winfree and Strogatz 1983), in the formation of Bénard cells of convection in fluid

dynamics (Segal 1965), or in populations in ecology (Yorke and Li 1975) cannot be

expressed or reduced to the movements or interactions of the individual units which supportand ultimately determine the higher level patterns. (See also van Gulick 1993.) Or consider 

the relationship between physics and biology. Many biological concepts only refer to the

higher level (with respect to physics). For example, many agree that "species" or "gene"cannot be defined in terms of lower level properties or entities because the definitions are

intimately tied to other properties and entities at the higher level (see in particular 

Fodor 1974, Kitcher 1984, Suppe 1989). I maintain that something analogous occurs

with higher level mental events and lower level physical ones.

In any event, we can now think of  causal interactions as operating on any level of 

organization, since no level is more fundamental. As a result, causality becomes a

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 15/100

 promiscuous relation. (This view of causality has been discussed rather extensively in

Sober 1984; see also Jackson and Pettit 1990.) As long as a particular occurance of  P (atany level) raises the probability for a particular occurance of Q (at any level) enough, we

have a  prima facie case for causality. Contrary to Kim's position, the ontological

dependency of the higher levels on the lower is now an issue orthogonal to the question

of causation. Insofar as we can define a higher level property, then that property mightbe causally efficacious.

The problem now is that we have too much causality, as it were. The Exclusion Principle

denies that no event may have more than one cause. Upon reflection, though, admitting

the possibility of causal overdetermination seems almost trivial. (See also the discussionin Humphries 1992, Yablo 1990). Suppose a three ton rock is lifted by a backhoe. Clearly

there is some causal connection between the movement of the backhoe and the movement

of the rock, and this connection exists because of some property of the backhoe in virtue of which it acted as a causal agent. What is this property? Is it that the backhoe exerted an

upward force greater than three tons? Is it that the backhoe exerted an upward force of 3.21

tons? Is it that the backhoe exerted an upward force greater than the downward force of therock? As Fodor remarks, "it may be that [c] . . . has many -- even many, many --

properties in virtue of which it is capable of being the cause of [ e] . . ., and it need not

 be obvious which one of these properties is the one in virtue of which it actually is thecause of [e] . . . . At least, I can assure you, it need not be obvious to me " (1989, p. 64).

(See also Heil and Mele 1991.)

In any event, this perspective does not entail that mental states (or other higher level

 phenomena) are superfluous and are candidates for elimination from our explanations.Instead, mental states are higher level physical properties that can enter into theories just

as easily and as frequently as any other regularity science currently sanctifies . At first pass,

this multi-layered view of the world overcomes the problem of mind by explicating how the

mind is physical (hence, potentially causal), yet not reducible to lower level descriptions inany interesting way. And it seems to give us what we want -- autonomous sciences and a

place for the mental in explanations as causally potent.

II

We now have arrived at a useful middle ground. Mental phenomena are just as good andreal and useful as any other higher level property, which are as good and real and useful as

any other lower level property. Our world exists as a nested hierarchy in which the different

levels of organization follow different -- perhaps fundamentally different -- patterns of  behavior and interaction. At each level though one may distinguish regularities in the

observed phenomena that are ultimately used to ground models and theories that describe

and predict the phenomena. Mental events then are just one more type of (higher level)

regularity that we observe. Lower level physical events are not explanatorily prior to thehigher levels, nor are they clearly metaphysically superior to them. 

However, there remains a gap between regular higher level properties and higher level

 properties causally responsible for some course of events , for not all properties are causal.More work must be done before we can answer when one level of organization contains

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 16/100

the objects and concomitant properties responsible for some event. Audi (1993) points

out, for example, that we might legitimately claim that Nixon was run out of office becausehe was morally corrupt. But that fact does not thereby entail that moral corruption quamoral property was causally responsible. Presumably, things like the Watergate

investigations, Nixon's duplicity, the sentiment of the American public, and so on, are the

sorts of things that actually caused Nixon to resign. (See also Fodor 1987, Jackson andPettit 1990 for more examples.) Perhaps more to our point, consider Pat staring at her 

sunburned nose in the mirror. It seems natural to claim that Pat's fair skin (among other 

things) is causally responsible for her reddened face; however, it is not responsible for her reflection's pink appearance (a reflected wave triplet is responsible for that). Nevertheless,

Pat's fair skin supervenes on both Pat herself and her reflection. Why should we think that

mental events resemble actual sunburns more than reflected ones? How do we know that

mental events are responsible for some behavioral effects instead of being some

illusory causal pretender?

Moreover, even though causal relations turn out to be relatively ubiquitous (from my

 perspective), only some causal relations can be privileged in explanation. Even thoughevents may be causally overdetermined, their explanations are not (see especially Fodor 

1974, Hardcastle 1991, Kitcher 1984, Putnam 1973). Different explanations of the same

event are not conceptually equivalent. They do not impart the same information. Whichinformation is relevant to the explanation at hand is the question.

On two counts then, we have to decide what sort of weight we should assign to the higher 

levels of organization in the brain or mind in explaining behavior. In particular, we need

to answer whether  there is an epistemologically privileged higher level of organization for the brain that we could legitimately use in our explanations of behavior and the like

(that is, whether mental states are appropriate explanans given materialism ) and how we

would know this.

Fodor (1989) answers by arguing that we can claim that mental properties are causallyresponsible if they are subsumed under causal laws. However, Fodor's recommendation

without further discussion or elaboration begs the question. We want to know (among

other things) whether the so-called intentional "laws" of our folk  (or Freudian)psychology or our psychology proper  are in fact legitimate causal laws. Are these

 perceived patterns in fact   grounded in some causal relationship, or are they parasitic on

something else?

But within the attempts to flesh out (llegar o hacer sustancial)this suggestion lies a greatcontroversy that goes back to Davidson's "anomalous monism" (1980) (or at least LePore's

and Loewer's reconstruction of Davidson). Donald Davidson (and others, e.g., Kim 1983,

1992a, b; LePore and Loewer 1987, 1989) holds that only strict or basic or exceptionless

laws support causal interactions. That is, only things like the laws of particle physics can provide the nomologically sufficient conditions for true causal transactions. Psychological

laws, even psychophysical laws, are neither strict, nor basic, nor exceptionless. They are

used to relate sets of events only in conjunction with ceteris paribus clauses. Hence,they cannot support true causal interactions. Hence, mental properties qua mental are

causally suspect (anomalous).

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 17/100

In response, Fodor (1989) denies that "hedged laws can't ground mental causes" (p. 72). We

can see that ceteris paribus laws can do "serious scientific business" (p. 73) (where a

ceteris paribus law has the form: " M  -- >  N ceteris paribus") because they hold

nomologically or strictly whenever the ceteris paribus conditions are satisfied. We can see

this because we can distinguish meaningfully between the claims " M 's cause  N 's ceteris

 paribus," and " M 's cause N 's, except for when they don't." The real difference between themore basic laws and the laws in the special sciences is that the latter require an additional

or a mediating mechanism in order to implement the causal transaction, while the former do

not.

But this debate over  ceteris paribus conditions is a red herring (pista falsa) because, in point of fact, all theories rely on ceteris paribus hedges. The real difficulty is that Fodor,

LePore, and Loewer are relying upon a deficient notion of scientific explanation. But if we

change our notion of scientific theories a bit, the debate disappears. Indeed, with a differentunderstanding of laws and explanation, we shall see that the answer whether we can use

mental phenomena in explanations turns on (conecta) how to understand the relationship of 

causality itself within a model-theoretic framework.

It is simply a fundamental error to assume that the basic laws of physics (or any laws of anyscience) are not hedged in important and fundamental ways. Scientists have no intention of 

accounting for all the intricacies of their subject matter. Instead, they abstract away a small

number of parameters from some collection of observed phenomena in order to form a data set , which is then used to explain and predict something (though not everything) about the

 phenomena themselves. The "simplifying assumptions" of physics amount to (llega a algo)

the same thing as the "hedges" of psychology. To take a clear example from "basicscience," classical particle mechanics uses only mass, velocity, distance traveled over time,

etc., in characterizing falling bodies. Moreover, it does not concern itself with the actual 

velocities, but only with velocities that would occur with an extensionless object in a

frictionless environment. Particle mechanics predicts "behaviors" that depend only upon the position and momenta of extensionless points interacting in a vacuum.

Similarly, the genetic theory of natural selection characterizes evolutionary phenomena in

terms of changes in the distributions of genotypes across a population over time as afunction of the rate of reproduction, the frequency of crossover, and so on. Behaviorist

theories in psychology describe the behavior of idealized organisms as a function of 

stimulus-response patterns and reinforcement schedules. Scientific theories describe the

 behavior of abstract mechanisms under ideal conditions which only approximates the behavior of real phenomena in virtue of a few "fundamental" properties under  normal

conditions. Following Suppe (1989), we can call this imaginary domain a  physical  

 system, an abstract replica or model of the phenomena that characterizes how the phenomena would have behaved had the idealized conditions obtained.

Science, then, does not apply its laws directly to observed or hypothesized phenomena, but

rather uses laws to explain the behavior of physical systems abstracted from the phenomena

in such a way that the behavior can be correlated with the phenomena (see also Bogen andWoodward 1988). A successful correlation then leads scientists to identify the attributes of 

the physical system with properties in the real world, thereby explaining them in terms of 

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 18/100

the posited laws. What we actually observe must be stretched and pruned such that we can

talk about what we would have observed if only the few relevant parameters of the phenomena existed under ideal conditions. These statements of altered observations are

used with the theory to make predictions about the physical system. The predictions are

then converted into statements about real phenomena by just reversing the procedure for 

altering the original observation statements.

Of course, what I have articulated is nothing more than a version of the semantic view of 

theories (cf ., Bickle 1993, Suppe 1989, van Fraassen 1980). In general, the semantic view

holds theories to be abstract pictures of relations among variables. A picture with all of its

 parameters fixed is a model. Each theory has several (maybe indefinitely many) models,and we can think of a model as representing one of the possible worlds allowed by the

theory. These models range in the amount of empirical or semantic content, from highly

abstract, purely formal models to detailed sets of observed phenomena and their relations.What links the models together as a single theory then is their common mathematical

structure.

What is important for our discussion about ceteris paribus clauses are how semantic

conceptions (and related views) define the scope of laws. Theories are interpreted to makeuniversal statements about the set of possible objects which fall under the theories. Theories

involving mental properties might encompass things like living creatures, or cordate

organisms, or mammals, or humans and primates, or just  Homo sapiens. Regardless, aslong as these sets are describable by lists of natural attributes, they can be subsumed by

 proper causal laws in some physical system.

By restricting the scope of laws to the domains of the theories, the notion of a law that is

"universally applicable" assumes the same meaning for laws both in the hard sciences andin the "soft" ones. Hence, the laws in psychology and the other special sciences would be

no different from any more "basic" law of particle physics or QM. The domain of physics

might be larger (indeed, it is supposed to cover all systems in this world); however, thelaws of physics are not thereby different. The scope of psychology is merely smaller, but its

laws are just as universally applicable in its smaller domain as the laws of physics are

applicable in its larger domain.

All sciences prune their observations to conform to simplified and idealized circumstances;they then all implicitly hedge their theories by making their laws concern only the abstract

 physical systems. Psychological theories and the theories in physics are both devised to

cover abstract physical systems, pared down (reducir) and idealized versions of someaspect of the world. As a result, they both operate two removes from actual phenomena and

 both focus on correlating sets of data to build models. Their corresponding laws are

sensitive only to these correlations, which may or may not correspond to how the world

actually functions in all its messy (desordenado) details. Hence, we can see that whether and how ceteris paribus clauses or additional mediating mechanisms are wielded (ejercer,

manejar) would be inconsequential.

But saying that psychological laws and theories are just like any other "more basic"physical law or theory is no more helpful than claiming that supervenient properties are

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 19/100

 just like the instantiating properties. It does not answer our problem. If we grant that

causation is everywhere, the question then becomes which causal chain we should be

using in our explanations.

Let us review what steps we have taken. We have moved from a metaphysical issue (Are

higher levels of organization causally efficacious?) to an epistemological one (Whichcausal history is relevant to the matter at hand?) in using a weak crieterion for being acause. Events might be causally overdetermined, but explanations trace only one causal 

 path. Which path to pick is not determined by whether the explanans refer to higher level

states or whether the explanation is hedged, for  I am assuming that higher levels of 

organization are just as legitimate as lower levels of organization and that all 

explanations actually refer to idealized sets of events. With the metaphysical perspective I

adumbrate, we naturally slide from ontological questions to epistemological issues. And

these issues I take to be most fundamental.

Here then is the difficulty: Suppose that idealized versions of c and e are subsumed by the

law  M  -- >  N . If we be materialists, then we would have to agree that for this particular instance of c -- > e, there is also a "more basic" law that likewise subsumes the events,

perhaps idealized differently. How do we determine which law is the appropriate one touse in an explanation of behavior? Saying that psychological laws are just like any other 

accepted scientific law is not enough, and we have seen that Kim's use of simplicity is not

telling. What would be?

Let me be clarify what I am asking. I grant Nelson Goodman's point that "regularities arewhere you find them, and you can find them anywhere" (1979, p. 82; see also Dennett

1991, van Gulick 1993). But I am assuming that in building a physical system, we pick 

out which regularity or regularities are especially salient.  How we pick out these patterns is a separate question and one that I presume turns on (conecta pone en

 juego)personal interest, the history of the inquiry, background theories, the shape

(influencia, condiciona) of the query (cuestionar, plantear una interrogante), and so on. Ileave that question aside. What I wish to focus upon is what happens after  the physical

systems are constructed and certain regularities highlighted. (interesante, destacado)

I want to know: If it is the case that more than one model can subsume each particular

c -- > e, then how do we know which model to use? (Or are all equally applicable?) Thisis not just a question of which model best captures some regularity. Because different

models may subsume c -- > e differently -- that is, because different models may emphasize

different aspects of the interaction and, as a result, may project their predicates differently-- which regularity we should be talking about  from the sets culled(sacrificado, 

 seleccionado) by science is the issue . Will mental properties end up being effectively 

epiphenomenal (relative to our accounts of behavior) in that models defined over those

sorts of properties should not be used in explanation?

III

In order to answer this query, we must first notice how the notion of cause is used in

model-theoretic accounts of science. I have suggested that science is, among other things,

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 20/100

engaged in correlating data sets. Physical systems are designed to capture the salient

 patterns such that we can then use them as a tool in predicting phenomena. If this is acorrect portrait, then a causal connection (in a model-theoretic version of science at least)

refers to nothing more than an interaction among the variables in a model.  If causes are

ubiquitous, then in virtue of highlighting an important regularity through

constructing a physical system, one has thereby pointed to a cause. 

What is important for our purposes is that, with this perspective, causal laws refer to

relations among variables in physical systems; hence they refer to the general types of 

events, objects, or properties, and not their individual tokens Moreover, causality becomes

 probabilistic. Causality becomes an expression of how likely one sort of event is to 

obtain, given that some other sort of event obtained . Since physical systems abstract

over many variables in the real world that might make a difference to particular outcomes,

the causal laws hypothesized to exist may likewise abstract over some factors relevant toindividual instantiations of the law. Hence, the best these laws can do is make a prediction

 based on incomplete information about how likely some event is. Nevertheless, we can still

use causal laws and presumed causal relations to make probabilistic predictions about howlikely e is to occur, given c (Pr(e|c)), or how likely  N is to be instantiated, given  M (Pr ( N |

 M )).

With the semantic interpretation of theories and a probabilistic notion of causality,

how to understand the relationship between mental events and their physical instantiations becomes one of how to understand the relationship between two models of the same set of 

events c -- > e. Certainly these models will generalize c differently, one in terms of 

property M , the other in terms of property P . They might even generalize e differently,one in terms of  N , the other in terms of Q. (Or there might be some generalization R neutral

 between the two models.) We can think of these two models as specifying Pr(e|c) as either 

Pr( N | M ) or Pr(Q| P ), where N and Q might be co-extensive. The question before us is: which

model presents a better explanation of  e -- the one that postulates Pr( N | M ) or the one that postulates Pr(Q| P )?

 Notice that now the question of how to understand mental causation is very similar to

more general questions in the philosophy of science (especially philosophy of biology)concerning how to distinguish truly explanatory relations from mere artifact. The accepted

wisdom there is that the relation of "screening off" (separar con un biombo) determines

which properties are the appropriate ones to use when framing a satisfactory explanation

(see, e.g., Cartwright 1979, Hardcastle 1991, Salmon 1971, Wimsatt 1984).

In general, we say that M screens off  P from R iff:

(1) Pr( R| P & M ) = Pr( R| M )

(2) Pr( R| P & M ) =/= Pr( R| P ).

That is, P adds nothing to M with respect to the instantiation of  R , but M adds something to

 P (see also Salmon 1971, p. 55). Screening off was originally designed to pick out the mostefficient cause of some event from some causal chain stretching over time. This would be a

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 21/100

"horizontal" use of the relation. We want the event that occurred closest in time to the

effect that is causally relevant to its appearance. We can then refer to that event as "the"cause of  e, thereby getting around (evitar llegar a ) Laplacianism, which claims that all

events in a causal chain going back to the first moment in time cause the present effect

under consideration. The most immediate cause presumably is the most interesting with

respect to explaining why e occurred.

However, I am proposing to use the screening off relation along a "vertical" dimension. In

this case, all the variables mentioned as potential causes occurred immediately prior to the

effect. Which cause is most efficient is not the issue. Instead, I want to know which of the

 purported efficient causes makes the most difference e, which is the most important. Tohelp clarify the connection between the probability calculations and predictability with

respect to vertical organization, consider the example of organismic selection. We say that

the phenotypic properties ( M ) screen off the genotypic properties ( P ) from reproductivesuccess ( R) because changing the phenotype without changing the genotype can affect

reproductive success (Brandon 1982, as discussed in Lloyd 1988). Similarly, when

discussing, say, the issue of levels of organization for physical systems, we might say thatthe higher level description  M of the informational patterns generated by individual firing

neurons screens off the lower level descriptions of the individual neurons  P  when

accounting for some behavior  R, just in case  R is strongly predictable from M alone, but Ris not as predictable from P alone.

What is important to notice is that the screening-off relation is asymmetric across

vertical levels of organization, while causality is not. That is, for each level of 

organization, we can point to an efficient cause of  e; however, not all of these causes areequally relevant to e's occurance . Determining which variables screen off and which

variables are screened off tells us which variables are more important. Merely isolating the

set of efficient causes across the levels of organization does not. Hence, screening off 

 presents itself as a likely candidate for a solution to the problem of artifact when choosingthe correct level of analysis in explanation. Is it likewise a plausible candidate for a solution

to the problem of mental causation in a material world? Is it the case that mental

descriptions screen off lower level physical descriptions, or vice versa?

Let us consider a simple example of pain. Suppose Terry has a severe headache and she

winces as a result. Without pretending that there are complete theories of pain in either 

 psychology or neuroscience, let us see how we might explain this phenomenon in terms of 

different levels of organization. We might say something to the effect that Terry'sexperience of pain caused her to recoil, or we might say something to the effect that

constricting blood vessels stimulates various nerves such that (among other things) her 

 brow furrows. Does pain screen off constricting blood vessels stimulating nerves, or viceversa?

The answer is not clear . Even though we are speaking of a particular instance of pain -- >

wince, how to understand the explanation of that particular event depends upon how we

would generalize the wincing event. If we take the wince to be a token of some reaction to pain, then constricting blood vessels add nothing to the explanation, since there are other 

means by which to have a headache. On the other hand, if  we take the wince to be a

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 22/100

reaction to a muscle contraction, then pain adds nothing to the explanation, since head

muscles can contract painlessly.

Which model to use in explanation does not admit of a principled answer. If we have areductive bias, then we might be inclined to generalize over muscle contractions and nerve

stimulations. On the other hand, if we lack that sort of bias, then we might be inclined totalk about pains and other sorts of mental events. And which sort of bias to have and whatkind of generalizations are best cannot be decided a priori. It all depends upon facts

 peculiar to the circumstance.

This is not to say that there is no way in which to make these sorts of decisions. Rather,

how we decide to describe Terry's wincing event depends upon explanatory biase(falsificado) s that are forged in the course of personal histories. It depends not only

upon which general questions we take ourselves to be answering, but also the context in

which the questions arise, the audience with whom we are conversing, other answers thatwe have accepted to similar questions, and so on.

Even though, say, psychologists and neuroscientists may have similar (sorts of) concerns

regarding Terry, the answers that neuroscientists give to their questions may not be of much

help to the psychologists, and vice versa. The difficulty lies in the pragmatics of 

explanation (see Hempel 1966, Kitcher 1989, Railton 1981, and especially van Fraassen

1980). Psychology and neuroscience operate in different contexts stemming from different

historical developments, which led to different investigative standards and methodologies,

different canons of knowledge, and different central questions (see Culp and Kitcher 1990)so that what counts as an acceptable answer to similar sorts of questions will differ.

We ask questions against a background of assumed facts and interests such that we want to

know why  R obtains instead of a particular alternative or set of alternatives (the contrastclass). Explanations tell us both why R is the case and why the contrast class is false. The

investigative context of a particular scientific domain determines which contrast class

is relevant in posing the problem in the first place . Whether something counts as an

explanation depends both upon the audience the scientists are addressing and the background of the inquiry. In this way, explanations turn on a tertiary relation among

 possible theories, known and accepted data, and historical context. So then, if 

neuroscientific explanations are going to compete with psychological ones with respect to aset of questions, the contrast class for the two domains questions would have to correspond.

But because psychology and neuroscience operate in such different academic

environments,  prima facie it is doubtful that the contrast classes do correspond. Atleast, we would need an argument that the contrast classes are importantly similar.

How to explain something and what counts as a good explanation depend upon a whole

host (muchos) of factors beyond the philosophical purview of this paper. Much has been

written on this topic in modern philosophy of science (see, e.g., Kitcher 1984, 1993;Railton 1981; van Fraassen 1981, 1988) and I certainly have nothing new to add here.

Instead, as a conclusion, I shall focus on what this position is not. In contrasting my

 position with other views, the move to reconceive the problem of mental causation in

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 23/100

epistemic terms rather than metaphysical ones should be clearer, as well as the open-

endedness of this brand of scientific pragmatism.

First,  contra Dennett (1987, 1991), I need to emphasize that I adopt epistemological

positions that have few metaphysical implications. That is, it is not the case that if we

can legitimately and profitably generalize c -- > e in different ways, then there is no"deeper" fact of the matter (1991, p. 49). I agree that "there could be two different, butequally real, patterns discernible in the noisy world. The rival theorists would not even

agree on which parts of the world are pattern and which were noise . . . . The choice of the

 pattern would indeed be up to the observer, a matter to be decided on idiosyncratic

 pragmatic grounds" (p. 49). However, these sorts of disagreements carry no ontologicalweight such that there is an "issue to be settled" (p. 49). Given that our theories concern

abstract physical systems, it is plausible -- indeed, it is expected -- that we might entertain

fundamentally incompatible conceptions and explanations of the same event. Asking whatgrounds one might use to decide that issue confuses the epistemic project with a

metaphysical one. The question of what is really out there (or whether there is anything out

there at all) is no longer apropos. That is, I am claiming that there are many patternssubsuming the same event that are genuinely causal. These causal chains exist at

different levels of organization. As mentioned at the close of in section I, none is

ontologically superior or more real; all are on a par.

One might question whether this move denying metaphyscial priority to some patterns over others would not thereby commit me to some sort of skepticism about the ontological status

of the mental. I think that it does not (or at least that it doesn't in any interesting way).

Whether any model involving mental causation actually corresponds to the world (andin a way such that we could know this) is just a question which recapitulates the

realism/anti-realism debate. Put in those terms, we can ask: Does a pragmatics of 

explanation thereby commit one to anti-realism? I think not. Criteria for determining the

truth of a theory or model operate independently of how the theories were forged. We believe a theory to be true, I presume, if it is explanatorily adequate, comports well with

other accepted theories, is relatively simple, etc. These issues are clearly separate from the

 pragmatic.

Suppose that we have two incompatible models of some set of events which entail

incompatible properties or objects (even taking into account that models capture idealized

and abstract physical systems). Obviously, at most one of these could be true. If the

 pragmatics of explanation allows that both models are equally acceptable, perhaps it seemsthen that one cannot maintain a version of realism (even with a list of truth-conducive

criteria) on pain of incoherence. However, in the remarkable event that both models are

equally explanatory, comport equally well (or equally poorly) with other accepted models,and so on, all we should and can do is withhold believing that either model is true until

something tips the balance in favor of one model over the other. But this particular position

with respect to this particular case says nothing about whether one should be a realist or anti-realist tout court .

I notice in passing that quite often in science we deliberately use theories we know to be

false (for pragmatic reasons). Though Einsteinian relativity has replaced Newtonian

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 24/100

mechanics, Newton's "laws" are still used in solving elementary problems because they are

simpler to use and get the job done well enough. It might be that, as Churchland (1981),Churchland (1986), Dennett (1987), and Stich (1983) seem to believe, our use of 

 psychological explanations will turn out to be likewise prudential. These explanations are

actually false,  but we use them nonetheless because they are simpler and get the job done

well enough. I have no idea whether they are right; time will surely tell. But regardless,eliminativist stories such as these cut no ice with the general issue of mental causation.

Even if our  current  psychological explanations are incorrect, that says nothing about

whether the replacement theories might not be true, or about whether the new theories willquantify over mental states. Mental causation qua mental remains viable, even if we grant

the eliminativists their arguments.

Second, claiming that there can only be pragmatic reasons for choosing one type of 

explanation over another keeps the playing field entirely open with respect to which sorts of reasons one might give for different sorts of behavior. Here I distinguish myself from

Yablo (1992) and Dretske (1988), who both suggest that we use the mental level of 

explanation when we wish to distinguish meaningful behavior from mere molar bodilymovements (see also Fodor 1987, Owens 1987, Shiffer 1982). We can point to legitimate

higher level explanations that invoke the mental (pain) to explain "mere" bodily movements

(a wince), as well as to legitimate lower level explanations that use neurophysiologicalinteractions (neuronal firing patterns) to explain "intentional behavior" (the withdrawal

(retirada, abandono) behavior to unpleasant (desagradable) stimuli in Aplysia) (Lockery et 

al . 1990). It is simply misguided to confine "mental" patterns to a single level of 

organization or allow them to be invoked in a single type of explanation. Our world is toorich and we are too clever to be so confined.

Hence, we arrive at a conception of mental causation in which mental properties are

 just as real as any other, but also in which we have the conceptual apparatus that allows

us to decide when we should invoke those properties as explanatory, given our historical biases and investigative goals. In the course of developing this conception, we have moved

from an metaphysical project (Is the mental causally efficacious?) to an epistemological

one (When is it explanatory?). Nevertheless, progress has been made, for determining whenit is appropriate to invoke the mind as a causal factor underlies the worry regarding whether 

mentalistic explanations can be legitimate.

Valerie Gray Hardcastle

Department of Philosophy

Virginia Polytechnic Institute and State University

Endnotes

References

Achenstein, P. (1977) "The Causal Relation," Midwest Studies in Philosophy, 4: 368.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 25/100

Anscombe, E. (1975) "Causality and Extensionality," in E. Sosa (ed.) Causation and 

Conditionals, New York: Oxford, pp. 178-.

Armstrong, D. (1989) Universals. Boulder, Colorado: Westview Press.

Audi, R. (1993) "Mental Causation: Sustaining and Dynamic," in J. Heil and A. Mele (eds.) Mental Causation. Oxford: Clarendon Press, pp. 53-76.

Baker, L.R. (1993) "Metaphysics and Mental Causation," in J. Heil and A. Mele (eds.)

 Mental Causation. Oxford: Clarendon Press, pp. 77-96.

Bickle, J. (1993) "Connectionism, Eliminativism, and the Semantic View of Theories," Erkenntnis, 39: 359-382.

Block, N. (1980) "Introduction: What is Functionalism?" in N. Block (ed.)  Readings in the

 Philosophy of Psychology, Volume 1. Cambridge, Massachusetts: Harvard University Press,

 pp. 178-179.

Block, N. (1990) "Can the Mind Change the World?" in G. Boolos (ed.)  Meaning and 

 Method: Essays in Honor of Hilary Putnam . Cambridge: Cambridge University Press, pp.

137-170.

Bogen, J., and Woodward, J. (1988) "Saving the Phenomena,"  Philosophical Review,XCVII: 303-352.

Boyd, R. (1980) "Materialism without Reductionism: What Physicalism Does Not Entail,"

in N. Block (ed.)  Readings in the Philosophy of Psychology, Volume 1, Cambridge,

Massachusetts: Harvard University Press, pp. 67-106.

Brandon, R.N. (1982) "The Levels of Selection,"  PSA 1982, Volume 1. East Lansing,Michigan: The Philosophy of Science Association, pp. 315-323.

Broad, C.D. (1925) Mind and Its Place in Nature. London: Routledge & Kegan Paul.

Campbell, K. (1970) Mind and Body. New York: Macmillan.

Cartwright, N. 91979) "Causal Laws and Effective Strategies," Nous, 13: 419-437.

Churchland, P.M. (1981) "Eliminative Materialism and the Propositional Attitudes," Journal of Philosophy, 78: 67-90.

Churchland, P.S. (1986) Neurophilosophy. Cambridge, Massachusetts: The MIT Press.

Culp, S., and Kitcher, P.S. (1989) "Theory Structure and Theory Change in Contemporary

Molecular Biology," British Journal for the Philosophy of Science, 40: 459-483.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 26/100

Davidson, D. (1980) Essays on Actions and Events. Oxford: Clarendon Press.

Davidson, D. (1993) "Thinking Causes," in J. Heil and A. Mele (eds.)  Mental Causation.

Oxford: Clarendon Press, pp. 3-18.

Dennett, D. (1978) "Current Issues in the Philosophy of Mind,"  American Philosophical Quarterly, 15: 249-261.

Dennett, D. (1987) The Intentional Stance. Cambridge, Massachusetts: The MIT Pres.

Dennett, D. (1991) "Real Patterns," Journal of Philosophy, 88 27-51.

Dretske, F. (1988)  Explaining Behavior: Reasons in a World of Causes. Cambridge,Massachusetts: The MIT Press.

Dretske, F. (1989) "Reasons and Causes," Philosophical Perspectives, 3: 1-15.

Eells, E. (1988) "Probabilistic Causal Laws," in B. Skyrms and W.L. Harper (eds.)

Causation, Chance, and Credence, Volume 1. D. Reidel Publishing Company, pp. 109-133.

Feigl, H. (1970) "Mind-Body, Not a Pseudo-Problem," in C.V. Borst (ed.) The Mind-Brain

 Identity Theory. New York: St. Martin's Press, pp. 33-41.

Field, H. (1980) "Mental Representation," in N. Block (ed.)  Readings in the Philosophy of 

 Psychology, Volume 2. Cambridge, Massachusetts: Harvard University Press, pp. 78-114.

Fodor, J.A. (1974) "Special Sciences, or The Disunity of Science as a Working

Hypothesis," Synthese, 28: 97-115.

Fodor, J. A. (1981) Representations. Cambridge, Massachusetts: The MIT Press.

Fodor, J.A. (1987) Psychosemantics. Cambridge, Massachusetts: The MIT Press.

Fodor, J.A. (1989) "Making the Mind Matter More," Philosophical Topics, 17: 59-80.

Goldman, A. (1969) "The Compatibility of Mechanism and Purpose,"  Philosophical 

 Review, 78: 468-482.

Goldman, A. (1970)  A Theory of Human Action. Englewood Cliffs, New Jersey: PrenticeHall.

Hardcastle, V.G. (1991) "Partitions, Probabilistic Causal Laws, and Simpson's Paradox,"

Synthese, 86: 209-228.

Hardcastle, V.G. (1992) "Reduction, Explanatory Extension, and the Mind/Brain Sciences," Philosophy of Science, 59: 408-428.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 27/100

Heil, J., and Mele, A. (1991) "Mental Causes,"  American Philosophical Quarterly, 28: 61-

71.

Hellman, G., and Thompson, F. (1975) "Physicalism: Ontology, Determination,Reduction," Journal of Philosophy, 72: 551-564.

Hempel, C.G. (1966)  Philosophy of Natural Science. Englewood Cliffs, New Jersey:

Prentice Hall.

Heslow, G. (1976) "Discussion: Two Notes on the Probabilistic Approach to Causality," Philosophy of Science, 43: 290-292.

Honderich, T. (1983) "The Argument for Anomalous Monism, " Analysis, 42: 59-64.

Honderich, T. (1988)  Mind and Brain: A Theory of Determinism. Oxford: Oxford

University Press.

Horgan, T. (1989) "Mental Quasation," Philosophical Perspectives, 3: 47-76.

Humphries, P. (1992) "On 'Mental Causation' and Emergent Properties," ms.

Jackson, F., and Pettit, P. (1990) "Causation in the Philosophy of Mind,"  Philosophy and  Phenomenologial Research, 1(suppl.): 195-214.

Kim, J. (1979) "Causality, Identity, and Supervenience in the Mind-Body Problem,"

 Midwest Studies in Philosophy, 4: 31-49.

Kim, J. (1983) "Supervenience and Supervenient Causation," Southern Journal of  Philosophy, 22 (Suppl.): 54.

Kim, J. (1989a) "Mechanism, Purpose, and Explanatory Exclusion,"  Philosophical 

 Perspectives, 3: 77-108.

Kim, J. (1989b) "The Myth of Nonreductive Materialism,"  Proceedings and Addresses of the American Philosophical Association, 63: 31-47.

Kim, J. (1990) "Explanatory Exclusion and the Problem of Mental Causation," in E.

Villaneuva (ed.) Information, Semantics, and Epistemology. Oxford: Blackwell, pp. 36-56.

Kim, J. (1992a) "'Downward Causation' in Emergence and Nonreductive Physicalism," in

A. Beckerman, H. Flohr, and J. Kim (eds.)  Emergence or Reduction? Essays on the Prospects of Nonreductive Physicalism. New York: Walter de Gruyter, pp. 119-138.

Kim, J. (1992b) "Multiple Realization and the Metaphysics of Reduction,"  Philosophy and 

 Phenomenological Research, LII: 1-26.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 28/100

Kim, J. (1993a) "Can Supervenience and 'Non-Strict Laws' Save Anomolous Monism?," in

J. Heil and A. Mele (eds.) Mental Causation. Oxford: Clarendon Press, pp. 19-26.

Kim, J. (1993b) "The Non-Reductivist's Troubles with Mental Causation," in J. Heil and A.Mele (eds.) Mental Causation. Oxford: Clarendon Press, pp. 189-210.

Kim, J. (forthcoming) "Multiple Realization and the Metaphysics of Reduction."

Kitcher, P.S. (1984) "1953 and All That. A Tale of Two Sciences," Philosophy of Science.

Kitcher, P.S. (1989) "Explanatory Unification and the Causal Structure of the World," in

P.S. Kitcher and W. Salmon (eds.) Scientific Explanation: Minnesota Studies in the Philosophy of Science, Volume 13. Minneapolis: The University of Minnesota Press.

Kitcher, P.S. (1993) The Advancement of Science: Science without Legend, Objectivity

without Illusions. New York: Oxford University Press..

Kornblith, H. (forthcoming) "The Metaphysics of Irreducibility," Philosophical Studies.

LePore, E., and Loewer, B. (1987) "Mind Matters," Journal of Philosophy, 84: 630-642.

LePore, E., and Loewer, B. (1989) "More on Making the Mind Matter,"  Philosophical Topics, 17: 175-191.

Lewis, D. (1973) "Causation," Journal of Philosophy, 70: 556-567.

Lewis, D. (1983) "New Work for a Theory of Universals,"  Australasian Journal of 

 Philosophy, 61: 347-377.

Lloyd, E.A. (1988) The Structure and Confirmation of Evolutionary Theory. New York:

Greenwood Press.

Lockery, S., Fang, Y., and Sejnowski, T.J. (1990), "A Dynamic Neural Network Model of 

Sensorimotor Transformations in the Leech," Neural Computation, 2: 274-282.

Macdonald, C., and Macdonald, G. (1986) "Mental Causes and Explanation of Action," Philosophical Quarterly, 36: 145-158.

Mackie, J.L. (1979) "Mind, Brain, and Causation," Midwest Studies in Philosophy, 6: 19-29.

Malcolm, N. (1982) "The Conceivability of Mechanism," in G. Watson (ed.)  Free Will .Oxford: Oxford University Press, pp. 127-149.

McLaughlin, B. (1989) "Type Epiphenomenalism, Type Dualism, and the Causal Priority

of the Mental," Philosophical Perspectives, 3: 109-168.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 29/100

McLaughlin, B. (1993) "On Davidson's Response to the Charge of Epiphenomenalism," in

J. Heil and A. Mele (eds.) Mental Causation. Oxford: Clarendon Press, pp. 27-40.

Menzies, P. (1988) "Against Causal Reduction," Mind , 97: 551-574.

 Nelkin, N. (forthcoming) "Patterns."

Oppenheim, P., and Putnam, H. (1958) "Unity of Science as a Working Hypothesis," in H.

Feigl, M. Scriven, and G. Maxwell (eds.) Concepts, Theories, and the Mind-Body Problem

(Minnesota Studies in the Philosophy of Science, Volume 2), Minneapolis: University of Minnesota Press, pp. 3-36.

Otte, R. (1985) "Probabilistic Causality and Simpson's Paradox,"  Philosophy of Science,

52: 110-115.

Owens, D. (1989) "Disjunctive Laws," Analysis, 49: 197-202.

Owens, J. (1987) "In Defense of a Different Doppelganger," The Philosophical Review, 96:521-554.

Pereboom, D., and Kornblith, H. (1991) "The Metaphysics of Irreducibility,"  Philosophical 

Studies, 61: 131-151.

Post, J. (1987) The Faces of Existence: An Essay in Nonreductive Metaphysics. Ithaca, New

York: Cornell University Press.

Putnam, H. (1960) "Minds and Machines," in S. Hook (ed.)  Dimensions of Mind . New

York: New York University Press, pp. 138-164.

Putnam, H. (1967) "Psychological Predicates," in W.H. Captain and D.D. Merrill (eds.)

 Art, Mind, and Religion. Pittsburgh: Pittsburgh University Press, pp. 37-48.

Putnam, H. (1973) "Reductionism and the Nature of Psychology," Cognition, 2: 131-146.

Putnam, H. (1975) "The Mental Life of Some Machines," in his  Mind, Language, and 

 Reality. Cambridge: Cambridge University Press.

Quine, W.V. (1969) "Natural Kinds," Ontological Relativity and Other Essays. New York:

Columbia University Press.

Railton, P. (1981) "Probability, Explanation, and Information," Synthese, 48: 233-256.

Schiffer, S. (1982) "Intention Based Semantics," Notre Dame Journal of Formal Logic, 23:

Schiffer, S. (1989) Remnants of Meaning . Cambridge, Massachusetts: The MIT Press.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 30/100

Seager, W. (1991) "Disjunctive Laws and Supervenience," Analysis, 51: 93-98.

Searle, J. (1983) Intentionality. New York: Cambridge Press.

Segal, G., and Sober, E. (1991) "The Causal Efficacy of Content,"  Philosophical Studies,

63: 1-30.

Segel, L.A. (1965) "The Nonlinear Interaction of a Finite Number of Disturbances to a

Layer of Fluid Heated from Below," Journal of Field Mechanics, 21: 359.

Shapiro, L. (1994) "Indication, Cognitive Science, and Lego Naturalism," ms.

Skillen, A. (1984) "Mind and Matter: A Problem that Refuses Dissolution,"  Mind , 43: 514-526.

Smart, J.J.C. (1963) Philosophy and Scientific Realism. London: Routledge & Kegan Paul.

Smart, J.J.C. (1980) "Sensations and Brain Processes," in C.V. Borst (ed.) The Mind-Brain

 Identity Theory. New York: St. Martin's Press, pp. 54-66.

Sosa, E. (1984) "Mind-Body Interaction and Supervenient Causation,"  Midwest Studies in

 Philosophy, 9: 271-281.

Sosa, E. (1993) "Davidson's Thinking Causes," in J. Heil and A. Mele (eds.)  Mental 

Causation. Oxford: Clarendon Press, pp. 41-50.

Stich, S. (1983)  From Folk Psychology to Cognitive Science: The Case Against Belief ,

Cambridge, Massachusetts: The MIT Press.

Stich, S.P. (forthcoming) "Puritanical Naturalism," in K. Neander and I. Ravenscroft (eds.)

 Prospects for Intentionality, Working Papers in Philosophy, Volume 2., Research School of 

Social Science, Australian National University.

Stich, S.P., and Laurence, S. (forthcoming) "Intentionality and Naturalism," in K. Neander and I. Ravenscroft (eds.)  Prospects for Intentionality, Working Papers in Philosophy,

Volume 2., Research School of Social Science, Australian National University.

Stoutland, (1980) "Oblique Causation and Reasons for Action," Synthese, 43: 351-367.

Suppe, F. (1989) The Semantic Conception of Theories and Scientific Realism . Chicago:

University of Illinois Press.

van Fraassen, B. (1980) The Scientific Image. Oxford: Oxford University Press.

van Gulick, R. (1985) "Physicalism and the Subjectivity of the Mental,"  Philosophical 

Topics, 13: 51-70.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 31/100

van Gulick, R. (1992) "Nonreductive Materialism and Intertheoretic Constraint," in A.

Beckerman, N. Flohr, and J. Kim (eds)  Emergence or Reductions? Essays on the Prospect 

of Nonreductive Physicalism. Berlin: De Gruyer, pp. 157-179.

van Gulick, R. (1993) "Who's in Charge Here? And Who's Doing All the Work?" in J. Heil

and A. Mele (eds.) Mental Causation. Oxford: Clarendon Press, pp. 233-256.

Wimsatt, W. (1984) "Reductive Explanation: A Functional Account," in E. Sober (ed.)

Conceptual Issues in Evolutionary Biology, The MIT Press, Cambridge, Massachusetts, pp.477-508.

Wimsatt, W.C. (1976) "Reductionism, Levels of Organization, and the Mind-Body

Problem," in G. Globus, G. Maxwell, and I. Savodnik (eds.) Consciousness and the Brain.

 New York: Plenum Press.

Winfree, A.T., and Strogatz, S.H. (1983) "Singular Filaments Organize Chemical Waves in

Three Dimensions," Physica, 8D: 35.

Yablo, S. (1992) "Mental Causation," Philosophical Review, 101: 245-280.

Yorke, and Li, T-Y. (1975) "Period Three Implies Chaos,"  American Mathematical 

 Monthly, 82: 985-992.

Skepticism, Stroud and the Contextuality of Knowledge.” Philosophical Explorations 4.1

(January 2001): 2-16.

we've moved to  philosophy.uwaterloo.ca/MindDict . Please update any links and go there

for the latest version.

eliminativism - The view that, because mental states and properties are items posited by a

 protoscientific theory (called  folk psychology ), the science of the future is likely toconclude that entities such as beliefs, desires, and sensations do not exist. The alternate

most often offered is  physicalist  and the position is thus often called 'eliminative

materialism' .

To be published in Behavioral and Brain Sciences (in press)

Cambridge University Press 2003

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 32/100

 Below is the unedited, uncorrected final draft of a BBS target article that has beenaccepted for publication. This preprint has been prepared for potential commentators who

wish to nominate themselves for formal commentary invitation. Please DO NOT write acommentary until you receive a formal invitation. If you are invited to submit acommentary, a copyedited, corrected version of this paper will be posted.

What to Say to a Sceptical Metaphysician: A Defense Manual for Cognitive and Behavioral

Scientists

Professor Don Ross

School of Economics

University of Cape Town

Private bag,

Rondebosch 7701

South Africa

Email: [email protected]

URL: http://www.commerce.uct.ac.za/economics/staff/personalpages/dross/

Dr. David Spurrett

Philosophy

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 33/100

University of Natal

Durban 4041

South Africa

Email: [email protected]

URL: http://www.nu.ac.za/undphil/spurrett/

Abstract: A wave of recent work in metaphysics seeks to undermine the anti-reductionist,

functionalist consensus of the past few decades in cognitive science and philosophy of 

mind. That consensus apparently legitimated a focus on what systems do, withoutnecessarily and always requiring attention to the details of how systems are constituted. The

new metaphysical challenge contends that many states and processes referred to by

functionalist cognitive scientists are epiphenomenal. It further contends that the problemlies in functionalism itself, and, that to save the causal significance of mind, it is necessary

to re-embrace reductionism.

We argue that the prescribed return to reductionism would be disastrous for the cognitiveand behavioral sciences, requiring the dismantling of most existing achievements and

 placing intolerable restrictions on further work. However, this argument fails to answer themetaphysical challenge on its own terms. We meet that challenge by going on to argue that

the new metaphysical scepticism about functionalist cognitive science depends on reifyingtwo distinct notions of causality (one primarily scientific, the other metaphysical) then

equivocating between them. When the different notions of causality are properly

distinguished, it is clear that functionalism is in no serious philosophical trouble, and thatwe need not chose between reducing minds or finding them causally impotent. The

metaphysical challenge to functionalism relies, in particular, on a naïve and inaccurate

conception of the practice of physics, and the relationship between physics andmetaphysics.

Keywords: Mental Causation, Functionalism, Reductionism, Metaphysics, Explanation.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 34/100

1. Introduction

Philosophy progresses with a tide-like dynamic. Every wave, no matter how strong it seems

while rolling in, is followed by a backwash, often nearly as powerful. This makes philosophical development difficult to identify except in long retrospect. For scientists who

try to take philosophy seriously in their work, this is bound to be frustrating

Philosophers have been deeply involved in the development of cognitive science. Theclassical essays by Hilary Putnam (1963, 1967a, 1967b, 1975a) and David Lewis (1972)

that articulated and promoted functionalist understandings of mind are among the

foundational documents in the literature of the field. Among other things, they showed howand why the study of information processing as conducted in early AI could and should be

integrated with psychology more generally. And however far in sophistication the cognitivescience community has since moved from the narrowly computational models of the 1960sand 1970s, it is hard to see how it would have gotten where it is now without them. So

 philosophers do not exaggerate when they claim that their discipline has contributed crucial

 bricks to the edifice of contemporary cognitive and behavioral science.

By ‘functionalism’ we understand any position which assigns serious ontological status

to types of states or processes individuated by reference to what they do rather than what

they are made of  – that is by reference to their effects, rather than (necessarily) their 

constituents. Functionalism of this sort was never without its critics, of course. From our  perspective, eliminative materialists (e.g. Churchland 1981) have been the most important

of these, and their arguments with mainstream functionalists have been immensely helpful

in the effort to see how the neurosciences and robotics best integrate with the morerationalistic projects derived from AI. However, avowed eliminativists have always been a

fringe, playing against a relatively monolithic functionalist consensus. For most of the past

thirty years cognitive scientists could be assured that the main currents in the philosophy of 

mind, especially regarding causation and explanation, were running in a directionsympathetic to their activities. This has involved more than encouraging cheerleading,

amounting to something of working scientific value. It has helped to guide choices amongst

research directions by clarifying just where and how cognitive science might strive for serious integration with nearby research programs in, e.g., neuroscience and the physics of 

dynamical systems theory (see 3.1 and 4.2 below) without simply collapsing into them.

We regret to report, however, that the backwash has set in. Were a cognitive scientist tostroll into a typical discussion amongst the ‘purer’ philosophers of mind at a professional

seminar in 2003, she would find that functionalism is under siege in such settings. Instead,

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 35/100

‘new wave’ reductionism[i] is the horse on which increasing numbers of philosophers are

 placing their bets.

We are, of course, being melodramatic here, and deliberately so. Good philosophers arerightly cautious about changing their minds or investing in fads, and worthwhile philosophical activity is not best seen as a war of ‘isms’. Nevertheless, as philosophers we

are concerned by the rise of a new scholasticism in philosophy of mind that, in the stated

 pursuit of a return to ‘real’ metaphysics, threatens a loss of contact with empirical cognitivescience. Our aims in this paper are, first, to substantiate this concern (thereby justifying the

melodrama), and, second, to offer grounds for resisting the arguments that inspire it. We

think that metaphysics – ‘real,’ professionally done, metaphysics - is an important part of 

all science, including cognitive science. But we also think that what is recently being promoted under this banner is based on an unhealthy disregard for the actual practice of 

science, and that too little philosophical discussion of it shows adequate concern for this.

To the extent that some philosophers allow their discussions to drift away from relevance toand coherence with scientific activity, the short term course of cognitive science will not be

much affected. However, since we would deplore a situation in which the conversation

 between philosophers and cognitive scientists wound down into separated silos, we think that a corrective with two aims is in order. One aim is to address philosophers

themselves concerning the fundamental errors we diagnose in the new scholasticism.

The other is to provide cognitive scientists with a manual for answering philosophers who

try to convince them that there is something wrong with their metaphysics. After all, to theextent that cognitive scientists respond to philosophers by just shrugging and going off to

another room, the conversation winds down; to the extent that they argue back, it continues.

Our discussion is organized as follows: In (1.1) we review the standard arguments for functionalism in the special sciences, and offer an account of the rise of the functionalist

consensus. In (1.2) we briefly describe the recent threat to functionalism. In (2) this threat is

examined in greater detail, looking first (2.1) at an influential argument againstfunctionalism, and then (2.2) at the form of reductionism which increasing numbers of 

metaphysicians think is preferable. In (3) we say a little more about the argument in favour 

of functionalism from the perspective of the special sciences (3.1), and also argue that the

reductionism suggested by the metaphysicians would be disastrous for those sciences (3.2).Section (4) contains the metaphysical meat of this paper. In it we distinguish some different

ways of taking metaphysics seriously (4.1), consider the relationships between explanation

and causation (4.2), distinguish two senses of causation, which we argue Kim conflates(4.3) and clarify a number of considerations relating to the nature of physics, and the

relationship between physical science and the metaphysics of causation (4.4). In (5) we

review our argument and offer a conclusion.

1.1. Functionalism, Philosophy and the Behavioural Sciences.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 36/100

Functionalism has a strong claim to being part of the methodological and ontological

underpinning of any special science. By ‘special’ science we have in mind any science notconcerned with justifying, testing or extending the generalizations of fundamental physics,

and hence most science including (see section 4.4) most of physics. Functionalism offersone way in which special sciences can defend their significance against Rutherford’s claimthat ‘there is physics, and there is stamp collecting’ (Birks 1963). What potentially

distinguishes a special science from stamp collecting is that it is organised around a

distinctive taxonomy of phenomena and a set of processes at some level of abstraction from

fundamental physical processes, which are non-redundant, amenable to scientific treatment,and to which a fully realistic attitude can be justified.

The original considerations which led to the development of functionalism were, as it

happens, primarily drawn from issues in the sciences of behavior: a responsesimultaneously to a simplistic behaviorist equivalence of behavior and psychological state,

and to the apparent ‘chauvinism’ of expecting that the specific mechanisms which

accounted for psychological states in humans should be regarded as reductive explanationsof those states in general.

With respect to simplistic behaviorism, the case for functionalism runs as follows.

Traditional behaviorism identified mental states with dispositions to particular 

 behaviors, and hence expected that mental states – insofar as these were of scientificsignificance at all - could be read directly off surface behavior, [ii] so reference to behavior 

could, and should, replace reference to mental states. One objection to this program

 pointed out that if mental states can, as seems likely, interact with one another, then there

will be neither fixed nor simple pairings of mental states and dispositions to particular  behaviors. This line of thinking suggests a place for  intermediate causal  roles played by

(at least initially) unobservable states between ‘stimulus’ and ‘response’. Note that in the

first instance these hypothesised intermediate states are characterised extrinsically, by

reference to the difference they make to observable states and relations between those

states.[iii] At this stage, at least, it is possible to be agnostic about what it is that makes the

difference in question, even while being confident that some difference is being made. This

space for agnosticism about what plays the functional role in question relates closely to

the second, and for  present purposes more philosophically contentious (conflictivo),motivation for functionalism.

In this case the contrast is provided not by behaviorism but by reductionism. In the heyday

of type-type reductionism it was expected that particular types of special science stateswould pay their ontological and causal way by being reduced  to types of some science

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 37/100

closer to fundamental physics, applied to the same systems. So, perhaps, the biological

 properties of a system could be reduced to its chemical properties, and from there chemistrycould be reduced to physics. Classic statements of versions of this view include Oppenheim

and Putnam (1958) and Nagel (1961). So, perhaps, the mental state of being in pain would

turn out to be, or already was, reducible to the fact of having activated C-fibers (Smart

1959, Place 1956),

[iv]

in the same way that ‘temperature’ was supposedly reducible to meankinetic energy of molecules (Nagel 1961: 338-345). On this view the type ‘pain’ was to beconsidered reducible to the type ‘activated C-fibers’ when some bi-conditional bridge law

was found enabling statements about pain to be translated into statements about C-fibers

and vice versa. The other way of motivating functionalism is, then, to note that the

 proposed reduction is open to a charge of ‘chauvinism’ (Block 1980b: 270), because even if the biconditional linking pain and C-fiber activity in humans held, there presumably could

 be, or already were, agents physiologically different to humans than nonetheless

experienced pain. So, in what came to be the standard jargon, even if what ‘realized’ painin people was something involving C-fibers, the ‘role’ of pain could be realized in different

ways in other types of agent.[v] This, in a nutshell, is the multiple realization argument

against type-type reductionism for psychological states, and by implication an argument for a science of psychology that spans (abarcar, atravesar) differences in realization.

The multiple realization argument can be deployed in various ways as a positive argument

for the functionalist project. In the hands of, for example, Fodor (1974, 1975, Block and

Fodor 1972) it is used to make clear that many, at least, of the special sciences areconcerned with entities and processes which are to some degree abstracted from the details

of physical realization. As noted above, practitioners of those sciences can afford

(permitirse) to be agnostic  about the physical details that realize the relevant kinds and

 processes, because the distinctive descriptive and explanatory contribution made by their work depends for the most part on extrinsic, functional, relationships between role

 properties. A simple and classic illustrative example of the argument here is Fodor’s

treatment of the notion of a mousetrap (Fodor 1968), conceived in functionalist terms as adevice which takes as input a live mouse, and produces as output a dead one. Clearly a

wide range of devices and designs are capable of realizing the mousetrap role.

The immediately preceding discussion has referred to the concepts of  roles and realizers instating and partly defending functionalism. This distinction (which is stated and clarified in

slightly different terms in Block 1980a) also allows two importantly different ways of being

a functionalist. The difference turns on whether one is inclined to identify the functional

states with the role they play , or with what the realizer of that role is in a given case . Put

another way, a functionalist might think that ‘pain’ or ‘money’ pick out either  the property

of having some other (physical) property which realizes pain or money, or that, properly

analysed, they pick out C-fibers firing or dollar bills.[vi] Saying that the description of the

functional state picks out the role indicates commitment to the view that even though their realizers could be very different, humans and, say, Martians could be in the  same mental

state when in pain. On the other hand, tying the function to the realizer entails that humans

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 38/100

in pain and Martians in pain are in different states, perhaps different types of pain, just

because the realizers of the roles in each case are different.

 Note that both flavours of functionalism are entirely compatible with materialism, or  physicalism. A physicalist functionalist of either type will be committed to the

principle that physics is complete, or causally closed, i.e., that there are no non-

physical, e.g. vital, fundamental forces (Papineau 1993, Spurrett & Papineau 1999,Spurrett 2001b). Similarly, she will be committed to the thesis that if you fix all of the

 physical facts, then you’ve fixed all the empirical facts that there are. Often, although not

necessarily, this aspect of functionalist thinking is marked by saying that functionalistsaccept supervenience  – the idea that there are no changes without physical changes.

 Note also that there is a genuine tension between the different ways of being a functionalist.From the perspective of realizer functionalism, the role variety rides roughshod over notiene la menor consideración distinctions which need to be taken seriously. So the

‘equivalence’ of a hundred dollar bill, a cheque for the same amount and a bag of coins

with a total value of a hundred dollars doesn’t amount to much when we have to try and say

something about why we can only use one of them in a vending machine, or why only oneof them can be bounced by a bank. On the other hand, from the perspective of  role

functionalism, too much attention to the realizers amounts to abandoning the apparent

unity of many apparently powerful and useful generalisations. Qua ‘money’, it has to begranted, there is a deep sense in which any realization of one hundred dollars just is the

same.

The importance of this distinction didn’t emerge immediately during the early articulationof functionalism. In the classic papers collected in Putnam (1975b), for example, role and

realizer versions are run together in a way that is, in critical retrospect, problematic.

Philosophers were quick enough to unearth the tension, however. By the 1980s, central

debates in the philosophy of mind revolved around arguments between role and realizer functionalists.[vii] However, for a number of years, up to the mid-1990s, the debates were

 preoccupied with the question of whether  semantic meanings , as bearers of functional

roles for beliefs, desires and other ‘propositional attitudes’, could or couldn’t be

individuated for the purposes of cognitive science just by reference to intrinsic properties

(causal, computational, constitutional or whatever), or were irreducibly relational. Thisrunning controversy was known among philosophers as the ‘internalism vs. externalism’

debate, and for awhile it seemed as if the dispute between realizer and role functionalists

turned mainly on it. Fortunately we need not describe its details here, because by the mid-1990s it was largely over, with the internalists – the believers in so-called ‘narrow content’

 – having mostly surrendered (see Fodor 1994, Ross 1997). At that point, some thought that

the philosophy of mind had made itself ready for thorough integration into cognitivescience. In particular, the strong connection between externalism about semantics and the

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 39/100

idea that narrowly computational models of thought need to be replaced or supplemented

 by more biological, environmentally situated and robotic ones (Brooks 1991), made the prospects for positive philosophical contributions to the scientific project look promising.

Some of that promise has been realized; we cite, for example, Clark (1997) and Rowlands

(1999) with approval in this connection. However, from these same years two ideas gained

strength among philosophers that encouraged scepticism about, instead of participation in,mainstream cognitive science. The first of these, the conviction that qualitative

consciousness is beyond the reach of functionalist method (Chalmers 1996 ), or, on

some formulations, any scientific method (McGinn 1991) is a manifestation of 

conservative metaphysics that we thoroughly deplore, for reasons given in Ross

(forthcoming) and Dennett (2001a, 2001b). This will not be our concern in the present

 paper, however. The second basis for metaphysical party-pooping, which is our presentsubject, encourages even deeper scepticism because it challenges not just functionalism’s

adequacy in a particular domain, but its coherence in general.

For reasons we explain in Section 2 below, realizer functionalism didn’t die with semantic

internalism. As far as we know, the first recognizably contemporary expression of theworry that states picked out by reference to functional roles alone can’t cause anything

is Fodor (1987). However, at that point the worry was deeply enmeshed in the internalist /

externalist controversy, so its subversive potential wasn’t clearly spotted. However, withthe passing of internalism it popped clearly into wide view. In Kim (1998) it finds book-

length and elegant expression,[viii] and our experience as casual anthropological observers of 

fellow philosophers indicates to us that the majority of philosophers of mind are, although

not unanimously persuaded by this version, inclined to take it very seriously and, if not

agreeing with its conclusion, to accept the same basic picture of how things are in science,especially physics, when engaging with it (e.g. Marcus 2001, Elder 2001). In what follows

we occasionally find (and cite) allies among ‘pure’ philosophers, and it is no part of our

project to argue that nobody should pursue these problems by primarily logical

methods.  Given, though, that we are here confronted with a piece of  metaphysics

claiming consequences for science, we take it as deserving evaluation with an eye to the

science and  the metaphysics. As Marras (2000) points out, and as we will explain, whatwas originally supposed to be a consideration against role functionalism but  for  realizer  functionalism now looms as a sceptical threat to all functional explanation in any science .

Our aim in this paper is to comprehensively respond to the basis for this scepticism, from

the perspective of behavioural and cognitive science. Doing this, however requires someexcursions deep into metaphysics. Some scientists will likely doubt that such excursions

could be worthwhile trips for them to go along on. Hearing that philosophers are making

themselves uneasy about the enterprise of cognitive science because of metaphysical itches,they may be inclined to respond pragmatically, saying “we feel fine, so  you stop

scratching!” Such responses, often heard when philosophers confront scientists with their 

metaphysical scruples, do not  just  express a macho attitude. It has been a widespreadopinion among philosophers of science for decades that philosophy has no privileged

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 40/100

epistemic perspective from which it legitimately can or should try to bend science to any

 prior ontological objective or methodology. We endorse this stance, in a fashion to be mademore precise shortly. However, we also agree with Kim (1998), that if metaphysics

matters then it had best be done seriously. We believe furthermore that metaphysics only

matters if it matters to science; and, finally, we believe, and argue below, that metaphysics

matters to science. Given all this, it of course follows that if the metaphysicalpresuppositions of cognitive science are causing genuine itches, then everyone ought to

care about scratching in the right place.

We find it necessary to say something about these grand themes for the following reason.Kim’s flagship argument against the recent externalist-functionalist near-consensus does 

have a scholastic aura about it; in particular, as philosophers take up Kim’s challenge and

gnaw away at the problems he has raised, they focus a great deal of their attention on subtledifferences amongst variations on the definition of the ‘supervenience’ relation. We’ll

ultimately conclude that this really is a scholastic’s response, in the bad sense of the word(if there is a good sense). But this generates two strategic concerns at the outset. First,

this conclusion may lead some philosophers to suppose that we are trying to have what Kimcalls a ‘free lunch’, that is, simply refusing to take the demands of metaphysics seriously.

Second, the fact that what we regard as a  scientifically interesting metaphysical problem

comes dressed in scholastic garb – it’s even based on something called ‘the supervenienceargument’ – will lead too many scientists to conclude right away that we’re engaged in an

in-house philosophers’ quibble that isn’t any proper business of theirs. These concerns

 present us with the following tactical burden. We must present the supervenience

argument, the basic grounds for the new disquiet, in a way that does logical justice to it

and  captures the gripping intuitions behind it that we don’t think you have to be

scholastically inclined to appreciate.

So, here goes. Our talk about `scientifically interesting metaphysics’ gestures at thefollowing fact. It is a feature of scientific epistemology, as really practised in laboratories

and journals, that the various pieces of scientific inquiry must broadly cohere into a general

world-view that, at least in its core, almost all signed-up members of the mainstreamscientific professions can share. Furthermore, it is a legitimate job of the ‘serious’

metaphysician to ensure that proposals for articulating and enriching this world-view are,

at least potentially, genuinely enlightening, and not merely verbal or technical. By‘genuinely enlightening’ we mean that such articulations should actually be able to help

scientists choose amongst theoretical and/or procedural alternatives in cases where the

empirical facts remain sufficiently underdetermined to leave options open in pragmatically

 pressing (as opposed to just logically possible) ways. Now, what we have just said is notvery precise, and so not very bold. But it is enough to help show why metaphysics can

be (and the issues raised by Kim’s supervenience argument are) scientifically

interesting. Our bland claim makes a minimal commitment to the idea that, at some levelof abstraction, the sciences need to ‘hang together.’ Ser coherentes However, this

commitment is in direct tension with the best motives for having special sciences, all of 

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 41/100

which turn on the facts that, along various dimensions both ontological and

epistemological, different sciences do not hang together, and that we’ll deny ourselvesimportant insights and generalizations if our respect for minimal metaphysics makes us

work too hard to try to get them to do so. Kim’s supervenience argument is  aimed

 precisely at this tension, and with unusually limpid clarity. Though, as we shall see, the

argument generalizes all the way across the sciences, it helps its clarity, but at the sametime especially challenges cognitive and behavioural scientists, that it is focused directly on

this tension as it arises within the domain of their work, which sits across the fault line

 between generalizing and special ontologies. So we think that a working cognitive scientistwho is confronted with Kim’s argument will and should then notice the tension every time

she goes to write up some new results, and will and should feel itchy. The problem then, we

will argue, is that Kim and other philosophers, instead of telling her where and how toscratch, counsel relief through professional suicide. We will be pleased to show that this is

not called for.

Kim’s argument is aimed directly at role functionalism. According to its conclusion, role

functionalism is not a stable metaphysical position. Instead, it collapses into a choice between epiphenomenalism and reductionism about mental properties, objects and

 processes. Kim assumes that epiphenomenalism would be a dire (extreme, horrendo)

outcome, both metaphysically and scientifically, but then spends much of his book trying tomake reductionism seem palatable. As we will show, he is not convincing. The

foundational assumptions of cognitive science, along with those of other special sciences,

deeply depend on role functionalism. Such functionalism is crucially supposed to deliver akind of causal understanding. Indeed, the very point of functionalism (on role or realizer 

versions) is to capture what is salient about what systems actually do, and how they

interact, without  having to get bogged down (quedarse atascado) in micro-scale

physical details. Functionalist understanding is, furthermore, supposed to deliver all the goods of  properly causal scientific work : permitting predictions, causal explanations,

sustaining counterfactuals, enabling the planning of interventions and so forth. But if 

reference to role properties can be shown to be causally redundant, as  Kim’s

argument purports to show, then the appearance of causal relevance is a sham, and

role functionalists, including most cognitive and behavioral scientists, most of the time, are

really only telling ‘just so’ stories to one another.

So, apologies for some coming scholasticism duly made, let’s now get this dangerous

supervenience argument onto the table.

2. The Armchair Strikes Back 

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 42/100

According to Kim, the key challenge to role functionalism turns on what he calls the

‘causal exclusion’ problem, which arises if he is correct that putative physical and

mental causes for the ‘same’ event can be shown to be in conflict. His problem, therefore,

is to provide an answer to the question: ‘Given that every physical event that has a cause

has a physical cause, how is a mental cause also possible?’ (1998: 38). This is the

 problem of ‘finding a place’ (1998: 2) for mind in a physical world, given the causalclosure of physics. The fact that Kim is concerned with the problem as a metaphysical 

challenge means that it won’t do simply to point out the pragmatic benefits, or 

indispensability, of  mentalistic explanations (including causal ones) without having agood metaphysical story to tell about how  and why  such explanations are legitimate (cf.

Marcus 2001). This would be the strategy, discussed and rejected above, of ignoring the

demands of metaphysics, asking, as Kim says, for a free lunch – keeping your comfortableintuitions by refusing to notice that they commit you to anything outside of cognitive

science.

2.1 Kim's ‘supervenience argument’.

Whether or not a cognitive scientist is in the habit of using the word ‘supervenience’,chances are good that some of her daily working assumptions involve at least a loose

version of the concept. Starting generally: one set of (e.g., mental) properties supervenes on

another (e.g. physical or neurobiological) set if, roughly, something cannot change withrespect to its supervening properties without undergoing some change with respect to its

subvening (‘base’) properties. Materialist functionalism involves commitment to

supervenience in this sense, insofar as it is reasonable to suppose that what role some entityrealizes cannot change without some physical changes taking place  somewhere. This

relationship of covariance plus some kind of dependence (because physical changes need

not lead to changes at the supervening level) is weaker than reduction, and does not commit

you to anything like realizer functionalism (let alone internalism) unless you add that therelevant physical change has to occur in the realizer .

Kim’s argument takes the form of a dilemma that ‘apparently leads to the conclusion that

mental causation is unintelligible’ (1998: 39). The dilemma has two horns: on one hornmind-body supervenience is allowed to fail, and in the other it is assumed to hold. For the

 purposes of formulating the dilemma Kim defines the mind-body supervenience thesis as

follows:

Mental properties supervene on physical properties in the sense that if something

instantiates any mental property  M at t , there is a physical base property  P  such that the

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 43/100

thing has  P at t , and [nomologically] necessarily anything with  P at a time has  M at that

time (1998: 39).

This definition is not perfectly general. Philosophers have generated a large literature thatdebates the merits and failings of alternative definitions of supervenience. What is at issue

in these arguments is the appropriate  scope to aim for in stating generalizations about

functional role-fillers. At least , `pain’ should apply generally and univocally to (most) people, and probably to creatures with which people share recent common (or, perhaps,

any) ancestors. Perhaps almost any life form would need a trip-wire system that alarmed it

 by making it feel bad. If so then Martians would have pain too, however different itsrealizers might be in them. Now, to help discipline arguments about this sort of thing, it’s a

useful strategy to first fix the essential  conditions on pain ; that way you hold your 

semantics fixed and can test the empirical facts independently. Philosophers fix essentialsemantics by considering various abstract possibility classes, or ‘possible worlds’.

Depending on how many of these classes you want to legislate supervenience relations as having to hold across, you get different logical definitions of the relation. Fortunately for 

our purposes here, Kim’s dilemma arises for any such definition, so we will treat theversion just quoted as exemplary.

Here’s the first horn of the alleged dilemma.

If mind-body supervenience, in general, were to fail, and we are committed to the causalclosure of physics, then it seems as though we could not make sense of mental causation.

Put another way, if the supervenience relation doesn’t hold, and mental causes do have

 physical effects then we’d have to deny the causal closure of physics – we’d be claiming a physical consequence of a non-physical cause. As materialists, or physicalists, we can’t do

this, so it looks like the supervenience relationship has to hold. (Kim takes commitment to

the causal closure of physics as being a ‘minimal’ requirement for physicalism). So far so

good – this is a standard motivation for endorsing supervenience if you aren’t willing to bea reductionist (e.g. Fodor 1987).

On, then, to the other horn.

‘Suppose that some instance of mental property M causes another mental property M* to beinstantiated’ (Kim 1998: 41). By the mind-body supervenience thesis,  M  has a physical

supervenience base P , and M* has a physical supervenience base  P*. Kim asks us to grant

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 44/100

(otorgar, ceder, conceder) that P causes P*. But, then, since M* is realised by P*, why have

an apparently separate causal claim to the effect that  M  caused  M*, especially when itseems as though once P , then M* was going to happen anyway? (following Kim 1998: 38-

56, see also Marras 2000). More precisely we seem to have to choose between:

 M* is instantiated on this occasion: (a) because, ex hypothesi,  M  caused  M* to be

instantiated; or (b) because  P*, the physical supervenience base of  M*, is instantiated onthis occasion (1998: 42).

Kim notes that the apparent tension above could be relieved by accepting that ‘ M caused

 M* by causing P*’. But, given that both M and M* have respective physical supervenience bases, we should ultimately grant that ‘ P  caused  P*, and  M  supervenes on  P  and  M*

supervenes on  P*’ so that the ‘ M -to- M* and  M -to- P* causal relations are only apparent ,arising out of a genuine causal process from P to P* ’ (1998: 45).

So: If you deny supervenience you seem to be abandoning materialism, which would be

terrible,[ix] and if you affirm it you get stuck with a choice between epiphenomenalism

about the mental, or reductionism. The former is an awful option for cognitive science.

Therefore, the only option is reductionism. This is genuinely amazing, since the very

point of  endorsing supervenience was originally to allow materialism without 

reductionism!

2.2 Kim’s reductionist proposal

Kim’s reductionism is not quite the standard (‘Nagelian type-type’) variety that people

still learn in undergraduate metaphysics and philosophy of science courses (See Marras2002). According to that model, you reduce some type  x to some type  y by justifying a

‘bridge-law’ to the effect that all of the causal and other law-like generalizations you can

state in terms of  x can be re-stated in terms of  y . (nageliana) Instead, Kim proposes areductionism that proceeds along the lines suggested by Armstrong (1981) and Lewis

(1980). The details of the proposal involve a crucial step called ‘functionalization’ which

involves “enhancing (realzar,aumentar)bridge laws … into identities” (Kim 1998: 97).[x] 

Identities, unlike bridge laws, give ontological simplification, and promise to explain why itis that the bridge laws hold true.  Functionalization is to be achieved by ‘priming’

(aplicación) the to-be-reduced mental property (the proverbial M ) for reduction, which

means reconstruing it in extrinsic or relational terms, i.e. specifying its causal

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 45/100

relations to other properties. So M is now ‘the property of having a property with such-

and-such causal potentials, and it turns out that physical property  P is exactly the propertythat fits the causal specification’ (1998: 98). It follows that  M  can be identified with  P ,

which would solve the causal exclusion problem, because one property cannot be in

competition with itself over causal relevance, and Kim thinks there is no problem about the

causal capacity of physical properties.

It is, of course, an open question to what extent such reductions are possible, and how

extensive, extends, important, amplia, the scope of any given functionalizing reduction will

 be. T he multiple realization argument (discussed above and, again, below) indicates thatfunctionally individuated properties can have very diverse realizers, so functionalising

reductions should be expected to involve some disintegration of the role properties. Kim

himself seems comfortable with this, describing the upshot of his arguments as beingthat ‘multiply realized properties are sundered (separados,cortados) into their diverse

realizers in different species and structures, and in different possible worlds’ (1998: 111).This is supposed to save  something  of functionalism, albeit at the expense of 

relinquishing (cede, renuncia) the capacity to say what it is that makes some apparentlysimilar functional properties related or the ‘same’ in cases where their  realizations are

significantly different. (We return to the question of just how much difference would count

as significant in due course.) Kim’s approach, interestingly, inverts the standard image of functionalism, traditionally regarded as a major form of antireductionism, since on his view

‘the functionalist conception of mental properties is required  for mind-body

reduction’ and is even ‘necessary and sufficient for reducibility’ (1998: 101). But is thisfunctionalism at all? Marras (2000) thinks not, and argues that Kim has ‘in fact given up

renuncia, on functionalism’ of which a central idea was that mental/functional properties

retained their ‘identity and projectibility across heterogeneous physical realizers’. Kim,

who claims to take multiple realizability ‘seriously’, concedes that to those who might

want to ‘hang on to’ functional properties as ‘unified and robust … in their own right’

his proposal will be a ‘disappointment’ but also maintains that the conclusion in question

is ‘inescapable’ (1998: 111).

 Notice at once that if there is any sort of functionalism still alive in Kim’s proposal, it’s

realizer functionalism, not role functionalism. So perhaps what Kim’s argument, and his

way out of it, shows is that if you want to try to be a  serious, anti-reductionist, functionalistthen you had, somehow, better be a role functionalist. As discussed in Section (1) above,

many have thought that since at least 1987; but initially the implausibility of semantic

internalism was the main reason. Now it turns out that there’s a more general reason: if you

try to be a realizer functionalist, you’ll turn ‘inescapably’ into a reductionist, and youwon’t be able to do cognitive science (or biology, or economics, or …)! Or so we now aim

to show. Remember, though, that showing we’d be in trouble if we followed Kim, no

matter how big the trouble, doesn’t show that we’re not in trouble. Acknowledging thatis the price of taking metaphysics seriously.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 46/100

3. Special Sciences without Functionalism

In section (1) of this paper we outlined the reasons for the establishment of a broadfunctionalist consensus in the behavioral sciences, and the special sciences more widely.

Functionalism seemed, was devised  to be, ideal for such sciences, insofar as it offered a

 justification for focussing on role properties and extrinsic relations, coupled with a well-motivated degree of agnosticism about the exact physical details of the systems studied.

In section (2), though, we described Kim’s supervenience argument, contending that

functional causal claims, understood as being claims about properties which supervene on

more basic physical properties, are epiphenomenal, and can only have their causal statussaved by reducing them to physical properties.

It is not essential that anyone view this as a problem. One simple way to avoid the

challenge Kim poses is to be an instrumentalist about functional claims. That meanscontending that metaphysical questions about the causal status of scientific claims just

aren’t important, and that what really matters is whether science is, in some sense or other,

‘useful’. It isn’t, after all, compulsory to worry about metaphysics. If you are indeed willingto say that, ultimately, the validity of some piece of science is determined on pragmatic

grounds, then this is your stop, and you can disembark right now. In so proceeding you are

allowing that you don’t mind if the behavioral sciences are considered to be a kind of stamp

collecting – a process of arranging the artifacts of our own epistemic limitations ininteresting or useful-seeming ways. (As we argue shortly, in so doing, whether you like it

or not, and more to the point whether he likes it or not, you’re agreeing with Kim, because

the only place he leaves open for the special sciences is an instrumentally justified one.)

If you’re still here then perhaps you want to be more than a stamp-collector. Perhaps you

want a defensible functionalist conception of ‘pain’ that generalizes across species, or of 

‘competition’ that generalizes across organisms and ecologies, or even of ‘mousetrap’which does justice to the varied assortment of gadgets you have around for the purposes of 

killing mice. In this section we aim to do two things: first (in section 3.1), extend section

(1) above by developing stronger and more sophisticated arguments against reductionism in

the special sciences; and second (in section 3.2), make clear that Kim’s proposal doesamount to turning special scientists into stamp collectors.

3.1 Explanation and Causation

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 47/100

It is a manifest fact about science that the various special sciences are partly constituted by

 parochial types of causal relations. Indeed, this is one of the principal things making them special . These relations are, furthermore, reciprocal functions of the accepted explanatory

schemata in the relevant sciences. This fact is, at least in the first place, sociological rather than metaphysical. One way of being what Kim derides as a free lunch seeker is to take thisas a brute fact in need of no explanation, supposing instead that the ‘specialness’ of each

special science, taken individually, is somehow self-justifying.  Part of what is involved in

heeding Kim’s enjoinder to take metaphysics seriously is acknowledging the need to say

something about the circumstances under which special science accounts are genuinely explanatory, where it is presumed that a genuine explanation is not merely something

 psychologically satisfying to someone, but must cite explanans that are both true and

informationally non-redundant. In this section, we will show that, in light of leadingaccounts of explanation from the philosophy of science literature, Kim’s version of 

reductionism would disqualify many or most  prima facie powerful special science

explanations.

Where special sciences are concerned, we can inquire about the explanatory value of a

specific account at either or both of two levels. An account might be genuinely

explanatory just relative to the particular ontological and causal structure of the science in

which it is embedded, but remain mysterious from the perspective of the wider standpointat which science as a whole is expected to ‘hang together’. Kim, of course, contends that

explanations citing mental causes have just this status unless we embrace his

reductionistic version of realizer functionalism. The inquirer who takes metaphysicsseriously seeks accounts of phenomena that are explanatory  both relative to the

ontological presuppositions of her special science , and to whatever wider  metaphysical 

 principles unite the sciences as a whole . The project of seeking explanatory generality of this sort is historically, actually and normatively,  part of the business of science. That is to

say that the naturalistically oriented metaphysics that we engage in is continuous with,

rather than separate from, what ‘scientists’ do. Our main criticism of Kim’s proposal, towhich we will devote Section 4, is that the particular wider metaphysical perspective he

takes for granted has no persuasive justification. At the moment, however, we are

concerned with tracing the consequences of Kim’s proposal for the special sciences, and

for the cognitive and behavioral sciences in particular. But since we contend that one suchconsequence would be the disqualification of a whole class of important (putatively, at

least) explanations, we cannot avoid some introduction at this point of general

considerations from the philosophy of science. For the moment, these considerations areintended to facilitate our discussion of special-science explanations. In Section (4), we

amplify them in a general treatment of the demands of serious metaphysics.

Kitcher (1976, 1981) has argued that ontological unification, either within a specialscience or across two or more special sciences, consists in the justification of common

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 48/100

argument patterns that hold within or across, respectively, the science(s) in question . This

claim is then substantiated through detailed analysis of the concept of an argument pattern,which is a set of ontological and structural primitives featuring recurrently in the

explanations given in the unified domain. Thus, for example, evolutionary biology is

unified by its recurrent use of explanations that cite measurable effects of environmental or 

other selection on the distribution of varying and heritable properties within populations. Abiologist does not, qua   biologist, query pregunta the cogency fuerza of this sort of 

explanation in general, since accepting the soundness of its generic logic and its general 

ontological appropriateness is part of what makes her a biologist . We need not hereendorse all the details of Kitcher’s analysis in agreeing that this idea identifies one plausible

element of the vector of (soft) constraints on explanatory unification. Over the course of 

his recently truncated career, the late Wesley Salmon explored another element of thevector, one lying more clearly and directly in the metaphysical tradition that seeks a basis

for ontological monism in one fundamental kind of ‘stuff’. That is, Salmon endeavored to

show something enlightening about the ontologies of all sciences by reference to general

micro-structural relations that bind all real objects and processes. In the philosophy of science literature, Kitcher’s and Salmon’s approaches are taken as offering rival bases for 

identifying good scientific explanations in the shared context of scientific realism. [xi] We

agree with Salmon’s (1990) view that while neither his approach nor Kitcher’s may furnish

a complete and ultimate analysis of explanation, they form a complementary pair of answers to a general question about what science wants and needs from philosophy of 

science.

In the context of our response to Kim here, we will be following a road that Kitcher and

Salmon have mapped quite explicitly in dialogue with one another. Kitcher   (1989)characterizes his work as analyzing ‘top-down’ explanation,  wherein we explain

phenomena by fixing their roles in wider ensembles of regularities, and he contrasts

this with ‘bottom-up’  (de detallar o de abajo hacia arrriba) explanation, the sort

analyzed by Salmon, which consists in identifying the causal-mechanical processes

that generate a phenomenon being explained.  Salmon (1990) endorses this idea of a

‘duality’ of explanatory approaches, which he takes to apply across the board. Thus, tocite one of Salmon’s examples, we provide a top-down explanation of industrial melanism

in peppered moths by means of the familiar story embedded in population genetics and

evolutionary ecology, and we would furnish a bottom-up account to supplement it if we

added facts about the synthesis of proteins that lead up to the production of differentlycolored wings.

As will be clear from our discussion in the previous sections, Kim can be happy enough

with this sort of duality in explanation. His difficulties turn on the fact that, according tohis analysis, top-down accounts of the Kitcherian sort cannot be causal. Neither Kitcher nor 

Salmon would necessarily disagree with this, since the duality they endorse is

epistemological rather than metaphysical. However, many typical explanations in the

behavioral and cognitive sciences seem to be simultaneously top-down and causal.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 49/100

Consider the following example, based on Hutchins (1995) which echoes many others

found in the current cognitive-science literature on intentional action. Some navigationsystems on large ships require two specialist ‘pelorus operators’, one on either side of the

ship, each reporting, with the aid of a special instrument (the pelorus), the angular position,or bearing, of visible landmarks. Pelorus operators do not select the landmarks, instead theyare specified by other members of the navigation team. Imagine a pelorus operator, recently

ordered to ‘stand by to mark’ the bearing of a particular landmark, and so expecting

immanently to be asked to report the continually changing bearing upon being ordered to

‘mark’. The actual response to the ‘mark’ instruction will be constituted by a series of neural, nervous and muscular events that the pelorus operator can’t directly access for 

description to himself, or subsequently report as distinct from one another (even if he

knows on theoretical grounds that they must have been).

His actions – including adjusting the orientation of the pelorus, maintaining a state of 

readiness to report the current bearing by frequently consulting the apparatus and what is

visible through it, and rehearsing and reporting the reading, will largely consist of pre- prepared subroutines that can be executed as relatively autonomous wholes. These

subroutines will be the product of training, guided by personal habits and primed by

ritualized social cues. Some subroutines will be specialized at gathering information from

the world (reading the instrument, decoding instructions about landmarks), some atcontrolling the information gatherers (lining the apparatus up on a landmark given an

external instruction), some at producing responses according to strict conventions

(reporting the bearing when instructed to ‘mark’, inter alia by producing the required phonemes in the required order). Others still will co-regulate the activity of those already

mentioned – preventing the reporting system from being executed until the ‘mark’

instruction has been decoded, etc. The routines will therefore partly be coded asdispositions in particular synaptic firing pathways, amenable to being triggered by some

small subset of those synapses.

Further, the pelorus operator’s entire brain must, on balance, be so configured that the

output of the instrument reading subroutines, when released by the decoded ‘mark’command, controls reporting behavior, preventing him from becoming enraged when

remembering the ‘Mark’ is also the name of a romantic rival, or abandoning his station to

tie a shoelace, and so on. He must instead be neurally primed to check and report the bearing at the moment of hearing the mark command, and do nothing else. So, there’s the

setup. And now, action! The command ‘mark’ is uttered and decoded, the visual position

relative to the calibrations on the instrument consulted, the markings transduced and processed, the result slotted into the conventional template, the phonemes rehearsed and

uttered ‘[Landmark X] 237’.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 50/100

This explanation is relentlessly (implacable) causal, but it is very far from strictly bottom-

up. The `subroutines’ to which we casually referred are black boxes, top-downcharacterizations of networks of connections that include both triggers and inhibitory links.

At every stage, we picked out these black boxes as pure role-fillers, by reference to a rich

conceptual network that we already know the operator must have learned. Perhaps, though,

we were just being lazy (avgo), or deferring (aplazar) to our own ignorance: if we couldhave provided the whole bottom-up story, individual electrochemical event by individual

event, wouldn’t we then have provided the exclusive causal story? Let us postpone that

question for now. Notice that even if a full specification of ordered synaptic potentials isthe exclusive causal story, then, as functionalists of all sorts have long emphasized, reciting

the specification would be a poor explanation of what happened, because there’s nothing

systematically special about these particular synaptic sequences that ties them to bearingreports from one occasion to the next. Furthermore, at one point we had to cite the

dispositional state of the pelorus operator’s entire brain! But this state will likely never 

occur again, exactly, no matter how long the operator’s career or how many bearings he

reports. And knowing the state in one case would do very little to illuminate different cases:what would those neurons, let alone (menos aun) the operator, have been doing were an

alternative landmark to have been specified? Or were the ‘mark’ command to have been

given a moment later? Thus the strictly synaptic account would miss almost all of thecounterfactuals relevant to behavioral explanation. The fundamental basis for this is the

servo systematic nature of the control architecture at work here. If some synaptic paths

wander away from the central task, then feedback generated from other regions concernedwith attentional focus will quickly recruit backup or alternative resources. Restricting

explanation to the actual microcausal chain misses this structural fact.

The account of the pelorus operator’s action given above is an instance of what Jackson

and Pettit (1988, 1990) call  program explanation. Social pressures operating on the pelorus operator ensure that one of many possible overall configurations of his brain that

keep him focused on his task will (likely) be in place as the moment for action looms. This

in turn ‘programs for’ one suitable chain of synaptic events or another, by virtue of thefeedback mechanisms through which brains embedded in environments control behavior in

general. Here is what Kim says about program explanation. First, he invokes one half of 

Salmon’s duality in asserting that “to explain an event is to provide some information aboutits causal history.” Then “what can be done is to define, say, the ‘causal network’ of an

event which is closed under both causal dependence and its converse, and then explain

the idea of explanation in terms of providing information about the causal network  in which an event is embedded . Pointing to an epiphenomenon of a true cause of an event

[thus] does give some causal information about the event” (Kim 1998, 76). In offering this

analysis, Kim does not disagree with Jackson and Pettit themselves. According to them, the

`programming for’ relation provides ‘causally relevant information’ but is not itself a causalrelation. That is, to them, knowing about the pelorus operator’s role plus the changing

position of the landmarks tells us that some causal process sufficient for a bearing of 

‘237’ being reported will unfold, but not which one.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 51/100

Suppose a scientist explains an animal’s hunting by saying that it’s hungry – in advance of 

knowing enough to have ‘sundered’ (romper separar) hunger by reduction into hunger lion,hunger mantis, hunger snake and so on. She would then be giving us a program explanation of the

hunting. Based on his remarks just quoted, Kim would concede that this explanation ‘gives

some causal information’. Furthermore, he seems to have no grounds for denying that it

gives the right causal information so far as prediction and generalization are concerned; for as the example of the pelorus operator is supposed to show, the program explanation

supports the relevant counterfactuals. So why are we supposed to still be worried about the

causal exclusion problem? Here’s why:

I believe it is only this sort of extremely relaxed, loose notion of  explanation that can

accommodate Jackson and Pettit’s program explanations. Explanation is a pretty loose

and elastic notion – essentially as loose and elastic as the underlying notions of understanding and making something intelligible  – and no one should legislate what

counts and doesn’t count as explanation, excepting only this, namely that when we speak of ‘causal explanation’ we should insist … that what is invoked as a cause really be acause of whatever it is that is being explained (Kim 1998: 76).

Implicit in this response is a metaphysical restriction on what sorts of states can and

cannot figure in ‘real’ causal explanations. Kim interprets minimal physicalism (that is,

commitment to the causal closure of the physical) as requiring that all properties that

cause things must be ( perhaps by reductive identification) physical properties. This, of 

course, invites us to ask  what makes a property ‘physical.’ Kim does not provide an

analysis, but merely a recursive restriction that ties the physical to the ‘micro.’ That is:“First, any entity aggregated out of physical entities is physical; second, any property that is

formed as micro-based properties in terms of entities and properties in the physical domain

is physical; third, any property defined as a second-order property over physical

properties is physical” (Kim 1998: 114-115). Then the idea is that as long as the domainof ‘real’ causal explanations is restricted to explanations that cite only micro-based

 properties, we are guaranteed never to violate the principle that physics is causally closed.

 Now we want to know what ‘micro-based’ means. Here is Kim’s definition of a micro- based property:

 P  is a micro-based property just in case  P  is the property of being completely

decomposable into nonoverlapping (supersosición, coincidentes )proper parts a1, a2, …, an,such that P 1(a1), P 2(a2), …, P n(an), and R(a1, …, an). (Kim 1998: 84)

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 52/100

Micro-based properties are thus macroproperties that are not shared by the micro-

constituents of the macro-systems that bear or instantiate them. So hunger lion could be amacroproperty, though hunger in general presumably couldn’t (see Section 3.3).

Thus on Kim’s view, whatever macroproperties ‘really cause’ molar behavior must be

decomposable into individual, nervous system–based, properties. This does not amount to

the absurd thesis that all causal powers at the macro-level are actually micro-properties ;Kim knows that cars can get people down the street while parts of cars can’t. Rather, what

he’s committed to is the thesis that a system’s causally effective macroproperties derive

their effectiveness entirely from interactions among causally effective micropropertiesthat are both regular and intrinsic to the same system.[xii] He answers worries about radical

multiple realizability of mental properties, with which both parts of this commitment areinconsistent, by suggesting that the possibility of practically interesting psychology shows

that, as a matter of fact, multiple realization is not out of hand:

The idea that psychology is physically realized is the idea that it is the physical propertiesof the realizers of psychological states that generate psychological regularities and

underlie psychological explanations. Given an extreme diversity, and heterogeneity of 

realization, it would no longer be interesting or worthwhile to look for neural realizers

of mental states for every human being at every moment of his / her existence. If 

psychology as a science were possible under these circumstances, that would be due to a

massive and miraculous set of coincidences (Kim 1998: 94-95).

Many cognitive scientists who see program explanations as playing ubiquitous andirreducible roles in their domain (along with those of other special sciences) do not agree

that it must be “the physical properties of the realizers of psychological states that generate

 psychological regularities and underlie psychological explanations.” Most will likelyconcede that similar neurophysiology (and other physiology) from one individual to the

next makes it possible for people to share comparable natural capacities, saliences and

learning histories, which is a necessary etiological condition for cultural learning.However, the operations of the natural devices that do this learning are not equivalent  to

our molar selves. Mental states are  individuated by a process of triangulating under  

equilibrating pressure from similarity of cogn itive and perceptual apparatus, similarity of 

social pressures on our histories of self-construction, and shared ecologies (especially socialecologies).[xiii] The basis for an interpretation of some set of synaptic potentials in the

pelorus operator’s brain as being ‘the state of believing that the bearing to the landmark 

was 237 degrees at the time he was ordered to mark’ is, in part , reference to his history as

someone conditioned to perform social roles, and, in particular, a role in a practice

that has such-and-such conventions.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 53/100

Explanations of this triangulating kind are pervasive enough across the behavioral sciences

that their genus constitutes a recognizable Kitcherian argument pattern. We identify

hunger-states by triangulating amongst physiological, ethological and evolutionary-

ecological factors; and then we furnish explanations of particular events in animal lives bysupposing that hunger programs for displays of search and consumption. We identify productive activities in economics by triangulating amongst considerations of energetic

output, behaviorally derived utility functions and culturally evolved rules of exchange; and

then we try to explain particular decisions of firms by supposing that production-possibility

frontiers and profit-maximization functions (given some cost of capital) program for theappearance at particular prices of goods on the market. There has been no shortage of 

attempts to rigorously ground this loose argument pattern of triangulation in a generic but

rigorous common logic – dynamic game theory, in which any of a variety of selectionmechanisms sifting amongst rival strategies for allocating scarce resources lead to

 predictable shifts in the distribution of behavioral tendencies, is the current favorite

candidate (Gintis 2000, Ross 2001). In these respects, the behavioral and cognitive scienceslook no obviously worse off, no intrinsically less unified as a suite, than the various wings

of physics and chemistry taken as a group. But Kim’s contention that special sciences are

only genuinely explanatory if they can survive a reductionistic re-interpretation does not depend on his finding that their typical explanatory attempts fail Kitcher’s criteria. Clearly,

for Kim, the unifying strategy championed by Salmon trumps Kitcher’s: the

epistemological duality is not mirrored at the ontological level. Scientists cannot

reasonably be expected to share this intuition, however, and throw away what look like powerful explanations from one leading philosophical perspective, unless serious,

 professional-class metaphysical arguments show that Salmon was more obviously holding

trumping aces than even Salmon himself thought.

We thus best press at the strength of the basis for  Kim’s `hyper-Salmonian’ intuitions

about explanation by asking first how they are supposed to make sense of actual

explanations in the behavioral and cognitive sciences, then, if that strains the prospects for accommodation, inquiring into the persuasiveness of their roots in general metaphysical

analysis by itself.

The second of these tasks is taken up in Section (4). In pursuit of the first question, let us

first note that the triangulational approach to the individuation of mental states in

psychology is compatible with two possible situations where the macro-micro relation

is concerned. On the one hand, mentalistic psychology and neurophysiology might employ

typologies that cross-classify across their putative micro-bases. In Kim’s words:

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 54/100

To say that a given taxonomic system cross-classifies another must mean something like

this: there are items that are classified in the same way, and cannot be distinguished, by thesecond taxonomy (that is, indiscernible in respect of properties recognized in this

taxonomy) but that are classified differently according to the first taxonomy (that is,

discernible in respect of properties recognized in that taxonomy), and perhaps vice versa.

That is, a taxonomy cross-classifies another just in case the former makes distinctions thatcannot be made by the latter (and perhaps also conversely). (Kim 1998: 68-69).

According to Kim, this amounts to a denial of supervenience as a one-way relation,

 permitting what Meyering (2000) calls ‘multiple supervenience’ (see below). Kim saysthat “this is a serious form of dualism, perhaps an approach worthy of serious

consideration”. Kim’s two uses of ‘serious’ here must prevent us from regarding this as

name-calling.(abusive lenguaje) On the other hand, we really don’t think that ‘dualism’ isquite the apt word here, since in this context it is clearly supposed to indicate views which

deny the causal closure of physics. We will indicate reasons for doubting thatacknowledgement of multiple supervenience implies such dualism, after first indicating

 just what multiple supervenience amounts to and why special sciences constantly traffic init.

3.2 Multiple Supervenience and Special-Science Explanation

Meyering (2000, 191) introduces the concept of multiple supervenience by means of ananalogy with dispositional explanation, and referring to the imagined example of Mary,

electrocuted while atop an aluminum ladder:[xiv]

… dispositions, just like macro-properties, fail to produce causal effects independently of 

their categorical base. And yet their explanatory power clearly differs from, or exceeds, that

of their bases. This becomes intelligible when we recognize that one and the same

categorical base ‘realizes’ more than one disposition. Even so, only one of those is

usually relevant for a given event. Thus Mary’s death is related to the electrical

conductivity of her aluminum ladder. But the categorical base thereof (the cloud of freeelectrons permeating the metal) also ‘realizes’ such diverse dispositions as the thermal

conductivity or the opacity of the metal.

The key point here is that the categorical base on its own, given that it realizes more than

one disposition, plays a less effective role in an explanation than does one particular 

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 55/100

disposition it realizes. Referring to the realizer is insufficiently precise compared to

citation of the disposition, or role. So the “actual realizer state is not merely inessential because a different state might have realized the same causal role. Rather it is inessential

 because the very same realizer state may yield a wide range of very different causal

trajectories” (2000: 193). (This nicely exemplifies why even Salmon came to recognize the

need for duality in explanation: Kitcherian top-down explanations are often moreinformative than  bottom-up ones, and objective informativeness is surely a 

metaphysically serious aspect of explanation on any reasonable account .) One way of 

describing the state of affairs Meyering considers is to say that there are supervenience

relations (i.e. relations of covariance plus dependence) going in two directions at once

here. On the one hand the disposition supervenes on a particular set of micro-properties, but

the disposition could be realized by different micro-arrangements. On the other , therelevant micro-properties realize multiple dispositions, and if a given disposition is picked

out in relational  terms, it turns out to supervene on the system of macro-relations. (The

earliest explicit appearance of this idea in the literature is Dennett (1981), who argues that

explanations in cognitive science often rely on ‘macroreductions’.)

If one acknowledges the possibility of multiple supervenience, then one disagrees with

Kim’s supposition that all supervenience relations point unidirectionally to physics.

This might suggest a basis for a quick answer to Kim’s supervenience argument, since if you reject its implicit premise that supervenience relations must all be ‘downward,’

then you won’t get impaled on the first horn of Kim’s dilemma (see 2.1 above), because

this kind of breakdown of supervenience has no consequences at all for the causal closureof the physical. To clarify this last claim, multiple supervenience does not imply the spooky

idea that you could change the global psychological state of the world while making no

 physical changes at all . But it does imply that, even given ideal science, you couldn’t

necessarily predict which  particular  physical changes would have to accompany a

given psychological change; i.e., that these relations aren’t, in general, systematic.

Avoiding Kim simply by abandoning supervenience, though, wins a cheap victory by

 burying more substantial issues at stake.(que estan en juego) Kim would presumably denythat an explanation citing, ascendente, upward-supervenient dispositions can be a causal 

explanation; and Jackson and Pettit, in shying away from regarding program explanations

as causal, presumably agree about this. This brings us to what we think is the deepest bedrock beneath the new metaphysical unease with the special sciences, with which we

grapple in Section (4).

Meanwhile, however, let us press on by asking what the special sciences actually do that

leads them to pick out entities, processes and kinds which don’t end up in neat( cuidado,orden) supervenience relations with physics. Meyering offers the following suggestion:

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 56/100

What gets studied in the special sciences is in fact huge systems of concatenated micro-

systems which are systematically organized in such a way that their typical causalantecedents prompt typical patterns of causal processing to eventuate in typical effects,

which in their turn serve as typical inputs for yet other causal sequences of events to take

 place. Regimented in this way the system produces emergent effects that have no

salience at the level of physics, and yet constitute the preconditions for the recurrenceof the sequence in question, or for the emergence of related processes which are

significant at that same level of special science description (2000: 193-4).

For an example of ‘emergent effects which have no salience at the level of physics’,

consider the huge collection of physical particulars which happen to constitute a given

stock market crash. Such an event is clearly of considerable importance to the group of 

special sciences we call economics. The claim being made here is that without  the perspective provided by the special science explanations in question there would be no way

of picking out that collection of particulars as being an event at all. It just wouldn’t be onanyone’s list of ‘things to be explained’, any more than the particular things counting as

‘money’ would cry out to be classified together on grounds recognizable to physics. So, itwould appear, if you want to have descriptions, let alone explanations, of phenomena

where functional, and especially multiple, supervenience obtains, then you need to grant the

irreducibility of the kinds which feature in such explanations. To be blunter still, we arefaced with a choice between embracing reductionism, or being able to construct the

explanations we do in fact construct.

Faced with this choice, some thinkers have supposed that there just can’t  be anythingwrong (no pude pasar algo?)with our apparently causal explanations, and hence that Kim

 just has to be wrong. One version of this response argues that if Kim is correct about the

mental causal exclusion problem (2.1 above), then all of the special sciences are in the

same trouble. Taking it as more or less self-evident that that can’t be the case, they reasonthat Kim’s problem isn’t a real problem at all. Burge (1993), Baker (1993) and Van Gulick 

(1993) all offer versions of this ‘generalization argument’. Kim’s response is twofold: he

argues that if the problem(exclusion causal) did  generalize, to reason that there isn’t a problem because we find the conclusion outrageous amounts to demanding a metaphysical

free lunch, and he argues that the problem does not, in fact, generalize very far.

If it seems like the causal exclusion problem should  generalize, it is because thesupervenience argument looks like it should apply to any non-physical property,

including chemical, geological, biological and other special science properties. In the limit,

this suggests that all  causation should ‘seep, filtrase, down’ to the level of micro-

physics. Kim  argues that this supposition trades  on vague intuitions about a hierarchy of ‘levels’ of properties, which need to be handled more rigorously. Specifically he argues that

we should distinguish between the realization relation and the macro-micro relation, and,

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 57/100

having done so, recognize that the ‘realization relation does not track the macro-micro

relation’ for the reason that both ‘ second-order properties[xv] and their first-order realizers

are properties of the same entities and systems’ (1998: 82). To supplement this argument

Kim develops a notion of a ‘micro-based’ property (discussed above in 3.1 and below in4.2) so as to save the physical status of ‘micro-based macroproperties’ such as hardness,

transparency, conductivity, and the like, as well as the objects in which we standardlylocate them such as tables, windows and nerves. Kim’s reflections here are, we think, partlysalutary: physicalism ‘need not be, and should not be, identified with micro-

physicalism’ (1998: 117). (Clapp (2001) develops independent arguments against

misleading ‘level’ talk, in the context of a defence of non-reductive physicalism.)

Marras (2000) argues, however, that Kim’s attempt to limit the extent to which the

causal exclusion problem generalizes is of limited success . At best Kim’s arguments

show that the causal exclusion problem is not an inter -level problem, indicating that the

only causation is micro-physical. What his arguments do not show is that it is not anintra-level problem for every individual special science. The possibility left open by Kim

is that every special science is ontologically confused, in virtue of classifying the world into

types which cannot be reduced to physics. In the light of what has been said above, itshould be clear that the causal exclusion problem generalizes, at least, to every case of 

multiple realization of a functional or relational property. (In a complementary contribution

Clapp (2001) shows that Kim’s argument has the ‘unsavory consequence’ that it makes all  

multiply realized properties, including  most paradigmatic physical ones, illegitimate partlybecause most properties are associated with causal/functional roles .) So the question how

many of the special sciences are threatened by Kim’s arguments is the question how many

of them trade in multiply realized functional kinds. We think that all of them do, but this isnot the place to defend this claim by means of an enumerative induction. A few examples,

then, will have to do a lot of work. Consider water.

On Kim’s view ‘being a water molecule’ is a straightforward physical property, whichhe regards as the ‘micro-based’ property of ‘having two hydrogen and one oxygen atom in

such-and-such a bonding relationship’ (1998: 84). This assertion is either false, or runs in

the face of the practice of chemistry. A sample of liquid water does not consist only of H 2O

monomer molecules, but also, at any moment, of various polymerous molecules such as(H2O)2, and (H2O)3, in a condition of statistical equilibrium involving rapid reciprocating

transformations (van Brakel 2000; Millero 2001, Ponce MS[xvi]). If we allow polymeric

forms of H2O to count as water, then water is multiply realized,   and Kim is simplywrong about what kind of property ‘being water’ is. Further, and more importantly, what

chemists recognize as procedures for determining sameness or heterogeneity of substance,

or establishing whether something is a pure element or a compound, are a variety of tests of 

which the most crucial involve attempts to separate a sample into its different constituents,and to determine whether it is hylotropic under phase shifts (Needham forthcoming, Ponce

MS). These procedures track relational, or dispositional properties – what it is that a sample

does rather than what exactly it is made o f. Following an account of these procedures Ponce

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 58/100

(MS) concludes that ‘chemical kinds are not, within chemical thermodynamics,

individuated by reference to their microstructure or micro-composition, but rather by

reference to their macroscopic physical properties, including their behavioral or

dispositional properties.’ Water is, perhaps, an especially telling example, just because if 

multiple realization operates at chemical scales, then it seems more likely to manifest at

larger scales, where the smaller scale variability could be inherited.

This is definitely what we see in cell biology,  where strict (Kim-style) reduction to

molecular biology seems impossible because key biological phenomena such as ‘signal

sequences’ are multiply realized and context dependent, and because functional rolesspecified in biological terms are indispensable. As Kincaid (1997) argues, many different

sequences of amino acids function as signals (multiple realization) but whether any given

sequence does so is partly dependent on context (since the same sequences in other contexts

don’t  play the signalling role – i.e. multiple supervenience), and, furthermore, ‘signal

sequences’ cannot be defined without reference to biological  functions. (See also Hull1972.)

 No matter how far Kim’s argument generalizes, though, we will not follow those who try tocall for a free lunch. We are simply after the interim conclusion that Kim’s problem, if it is

a problem at all, affects almost all of the special sciences. It could well seem as though

what is being argued for here is a kind of anthropocentrism, or pragmatism, where if something seems to us (or to chemists, biologists, etc.) like a good or powerful explanation,

then, whether or not it is amenable to being reduced, it should be regarded as legitimate.[xvii] Realism about special-science types does not require any such abandonment of 

metaphysical seriousness, however.

Macroscopic states need be neither anthropocentric nor pragmatically justified if there issome way of making sense of their being real, in the sense of ‘real’ which involves it not

 being up to us whether an ontology respecting Occam’s razor would have to recognize

them. Dennett (1991b), confronted with demands to take a position on whether ascriptionsof beliefs should be thought of in realist terms, or as merely instrumentally justified

devices, answered by offering a ‘mild realism’ in which the reality of basic physical states

was unproblematic, and in which macroscopic  patterns, understood in information- theoretic terms as structures that encode non-redundant, objective information  by 

means of compression, could be considered real enough to settle the debate. One of us

(Ross 2000) has argued elsewhere that Dennett’s position should be modified into a more

thoroughgoing pattern-realism, suggesting that a pattern should be considered objectivelyreal if and only if:

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 59/100

(i) it is projectible under at least one physically possible perspective

and

(ii) it encodes information about at least one structure of events or entities S  where thatencoding is more efficient, in information-theoretic terms, than the bit-map encoding of  S ,

and where for at least one of the physically possible perspectives under which the pattern is

 projectible, there exists an aspect of  S  which cannot be tracked unless the encoding isrecovered from the perspective in question (Ross 2000).

So considered, it is a contingent and empirical matter whether any particular real pattern isreducible to another, and, crucially, the question of the reality of any pattern is not to bedecided on anthropocentric grounds. This is so because patterns are required to be

 projectible under a  physically possible perspective, rather than a perspective which is an

artifact of human perceptual or cognitive capacities, so if there is a physically possible

 perspective from which some phenomenon recognized by our current working ontologycould be more efficiently represented under an alternative ontology , then our current

ontology is false, regardless of whether we are or are not, or shall ever be, aware of the

existence of the alternative possible perspective in question.

What realist special scientists do on this view, then, is seek to find real patterns in particular 

domains of reality, domains defined by sets of particular structures and/or processes at

some level of abstraction from fundamental physics. These patterns are what Meyering

needs to cash out his talk of ‘huge systems of concatenated micro-systems which are

systematically organized in such a way that their typical causal antecedents prompt typical

 patterns of causal processing to eventuate in typical effects’.

A defender of Kim’s line can object that what we have just said about explanation, and theirreducibility, indeed even the objective reality, of irreducible functional properties,

doesn’t automatically make any headway against the causal exclusion problem. It is,

after all, in the name (fundamental) of solving that problem that we are supposed to ‘giveup’ (renunciar, rendirse) on these irreducible properties. That is, it is just these properties

that we are supposed to learn ‘to live without’ so as (para) to preserve a coherent and

univocal concept of causation. Looked at this way, our banging the table and complainingabout how difficult it would be to live without the properties isn’t a good an answer to Kim

at all.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 60/100

It is true that the possibility of non-reductive realism about special-science types doesn’t

make direct  headwa y, because it doesn’t yet say anything yet about how to show thatspecial science generalizations invoking irreducible properties could be really causal . But

it’s more than a mere request for a free lunch, since it is crucial to showing what’s at stakefor the special sciences in evaluating the importance of Kim’s argument. We maintain thatKim’s position is based on serious misunderstandings about how things are in the

special sciences, and in order to make our more direct argument against him we need to

outline and defend what we take to be a more defensible picture. Because of his inaccurate

 picture of special sciences, Kim doesn’t seem to think the costs of his proposal areintolerable. We aim to show that they are utterly intolerable, requiring that we regard

almost all explanatory activity in the special sciences as confused.

3.3 Stamp collecting

In section (2.2) above we briefly outlined Kim’s reductionist proposal, which he urges as

the proper response to his supervenience argument for the instability of non-reductive

 physicalism. His proposal involves ‘sundering’ (separar)the types referred to in specialscience explanations in accordance with the particular reductive bases for them we discover 

empirically. We argue now that this effectively urges(impulsa) us to abandon functionalism

entirely, which goes against Kim’s claim to the effect that his brand of reductionism is

consistent with taking multiple realizability ‘seriously’ (1998: 111).

Here, to recap, is why Kim’s proposal is supposed to include elements of functionalism.

The process of reduction  he describes gets started with a role property  (pain, say) and

 proceeds via the discovery of the particular physical realizers  of that property to a series of reductive identifications, ‘sundering’ the role into as many realizers as turn out to be 

empirically warranted. One immediate difficulty here is that without access to the role

properties scientists  wouldn’t know where to start looking for realizers, or what therealizers were supposed to be realizers of . That is, as we argued above, if they started from

physical particulars and were prohibited from making reference to role-properties ,

it’s not clear that there would be any way at all for them to tell a collection of particulars

that was the realizer of a functional property from one that wasn’t, or to tell what manner of functional property it realized. This would be mission impossible: trying to look at some

huge mass of physical detail, and hoping to be able to say at some point ‘Ah ha! It’s a stock 

market crash,(accidents colapsa) and it started at this moment, and the proper boundaries

of the physical event constituting the crash are here.’ Kim’s proposal, that is to say,

requires that his metaphysically justifiable types of science are parasitic on the very types

he argues are epiphenomenal.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 61/100

A defender of Kim’s position may point out here that Kim does allow that by “grouping

 properties that share features of interest to us” it is possible that “important conceptual andepistemic needs” could be served (1998: 110). Perhaps, then, what we are calling

parasitism is what he would call serving an important conceptual need. This is anunsatisfying answer, though, since it leaves those hunting real causal relationships using

others whose work is epiphenomenal as trackers. It also makes clear, as we suggested

above, that on Kim’s view the only justification for functionally motivated special

science work is indeed instrumental: by doing that kind of science  you help the

reductionists figure out where to start digging, so as to dismantle the foundations of thatvery work. If we set this point aside and continue, matters only get worse for the special

sciences of Kim’s future world.

As we’ve seen, it is when empirical work turns up diversity in the realizers of somefunctional property that we’re supposed to dismember the role-property into its parts. Let’s

assume that Kim’s hunch,(presentimeinto) or hope, that realizers are likely to turn out to be

species specific is right – then perhaps we’d sunder pain, irrespective (sintener en cuenta)of how well it paid its way as a single notion in behavioral science, into  painh, painm, paino

(for, say, human, Martian, octopus). If we did this, we’d be proceeding as though we’d

discovered (so far) that pain was actually three things. How, though, would we decide

whether this was the case, or whether we’d really found out that there was no such thing as pain in general? Or, perhaps, that only one of the three was pain (in which case which

one?), and that the other two were something else? (See also Marras 2000, 2002.)

Looked at another way, had our  scientists somehow managed, despite the parasitismworry noted above, to  start  with a set of realizers (not having to work out from raw

 physical data what is a realizer of a function and what isn’t) it’s not at all obvious that they

would group the realizers in the same way as they would given access to the role propertiestoo. It could well be that, say, the realizer of Martian flatulence was structurally more like

the realizer of human pain than the two pain realizers were like one another. In this case

scientists working with only a collection of empirical descriptions of realizers might be

expected to group the realizers quite differently, if they were to group them at all. Therewould be nothing to stop them supposing, like neoplatonist medical thinkers, that walnuts

might be therapeutic for some brain conditions because they look rather like brains. Again,

it seems, reference to role properties by Kim’s rules has to be an instrumental necessityarising from the fact that the reductive relations are unknown at the outset of any enquiry.

Worse, the enquiry proceeds by making the role properties obsolete. If we call the three

imagined realizers of the pain role different versions of pain, it seems we are doing so outof a kind of nostalgia for when we thought (if we ever did) that pain was in some sense one 

thing, rather than out of clear-headed recognition of what, by Kim’s lights, we subsequently

discovered.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 62/100

The most important reason why the costs of going down Kim’s road are prohibitively

great is thus that it requires, in the end, giving up on the prospect of a unitary psychology, and in fact on any unified science referring to functionally individuated kinds.

(Economics and biology are obvious instances.) As noted, Kim is willing to allow that by“grouping properties that share features of interest to us” it is possible that “importantconceptual and epistemic needs”(necesariamente) could be served. But he is also adamant

that, in the end, functional properties with diverse realizers are properties “we will have to

learn to live without” (1998: 106). In other words the only justification for unitary sciences

having as their objects functionally individuated kinds is instrumental , because multiplyrealized properties turn out not to be metaphysically acceptable.

To hammer this point home, let us examine a real example. Consider hunger and satiety.

Hunger is multiply realized (perhaps, therefore, a property Kim thinks we may have to ‘livewithout’), by several mechanisms with distinct effects on different parts of the brain. We

can be stimulated to eat by, inter alia, the mechanical sensation of an empty stomach,

glucose level monitoring by the liver, the sight of others eating, the smell or taste of novelfood, and stress, not to mention combinations of these and other factors. One of the various

realizers of satiety, or of ‘contra-realizers’ of hunger, it seems, is hormonal, particularly

 but not exclusively via peptides occurring the in the gut. These  play a role both in

modulating other gut secretions, hence participating in the control of digestion, and insending an ‘enough’ signal to the brain. Whether these hormones do in fact realize one or 

the other of these functions, or neither, or both, at any given time, depends on relational

factors, so we here have a case of  multiple supervenience. They might be active, yet westop eating for other reasons (an artificially filled stomach triggered our mechanical

sensors), or continue eating despite their action, perhaps because people around us are

eating, or because we’re anxious, or because the novel dessert is more attractive than theunfinished main course.

A special science which studies what Meyering describes as ‘huge systems of concatenated

microsystems’ or what we have suggested should be thought of as real patterns, has a shot

at tracking typical patterns produced in consequence of the systematic organization of thosesystems. Such scientists get to make explanatory, and predictively powerful statements like:

When dietary variety is produced by providing a meal or diet composed of several foods,

animals generally become hyperphagic relative to single-food meals or diets (Raynor andEpstein 2001).

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 63/100

At the risk of laboring one point we have been emphasizing, it is worth drawing attention to

the term ‘variety’ in the above quotation. Pattern-hunting special scientists, such as behavioral scientists interested in motivated behavior, in this case eating behavior, are able

to justify broad-scope predictive generalizations referring to ‘variety’. In humans, the

variety in question is strikingly multi-modal, which is to say that the effect is stronger if the

foods differ in more than one way, including taste, color, shape, smell, texture, and presentation. ‘Variety’, though, just has to be multiply realized (there are different ways of 

 being different) and multiply supervening (structural features of the food may ground

various dispositions, only some of which contribute to ‘variety’ in a given context), and so by Kim’s lights it is one of those properties we’re going to have to learn to live without.

That, though, just cannot be acceptable. We hereby bet the farm that any possible life form

which metabolizes and is faced with resource scarcity will have  something , and in alllikelihood several things, playing the role of hunger, and that some of the generalizations

of, inter alia, our psychology, ecology, and micro-economics, will apply to it.

Kim doesn’t have a direct argument against multiple supervenience. It’s off his radar 

insofar as it is more powerfully anti-reductionist than anything he seems willing toconsider. We haven’t yet shown to be him wrong. What we have done is show just how bad

it would be for the special sciences were Kim’s position to be generally endorsed. Now we

need to look at how to disarm his argument that anything much needs to be changed in thespecial sciences at all.

4. Taking Metaphysics Seriously

We’ve stressed repeatedly that answering Kim requires taking metaphysics seriously. This

does not  mean respecting any particular  a priori hunch about the objects of any specialscience. Rather, it means acknowledging such demands on the structure of scientific inquiry

as transcend the disciplinary boundaries of individual special sciences, with the aim of 

 productively applying these demands to guide interpretations of the relationships amongsthypotheses generated across separate sciences. Part of our diagnosis of what’s wrong with

Kim’s approach is that he is mistaken about the relationship between one metaphysical

problem and the work of physics, so we begin this section in (4.1) by distinguishing anumber of metaphysical questions relevant to the issues at hand. In (4.2) we return to

multiple supervenience, and the related questions of what to count as physical, and how to

draw the macro-micro distinction in a way consistent with the account of realism offered in(3.2). In (4.3) we distinguish two ways in which ‘cause’ has been understood in the history

of philosophy, and argue that Kim equivocates between them. Finally in (4.4) we argue that

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 64/100

Kim erroneously supposes that physics provides us with the answer to a metaphysical

question, and furthermore that he is seriously mistaken about how things are with physics.

4.1 What metaphysics demands (or: How to pay for lunch)

Clearly the metaphysical question  bothering Kim is the following: What explains the

 fact that the supervenience relations which do in fact hold, hold at all? Kim thinks the

answer to that question would be a solution to the causal exclusion problem (2.1 above),and his own reductionist proposal (2.2) is supposed to show how the supervenience

relations hold because some stronger, reductive, relationship holds between physical facts

and functional (special science) facts: the causal capacity of special science properties can be inherited from the unproblematic causal capacities of the physical properties with which

we find they are identical. We agree with Kim that mere invocation of superveniencecannot answer the metaphysician’s question about the place of mind in a physical

world. If the special sciences that deal in supervenient types are not to be isolated from the

rest of our scientific ontology, we must indeed be able to explain why the particular 

supervenience relations (both general and specific) which in fact hold, hold at all.

As just indicated the causal exclusion problem, considered very generally, is a problemabout the unity of our scientific worldview, as briefly introduced in Section (1.2) above. In

the context of the naturalistic, broadly empiricist, conception of knowledge and reality

presupposed here, the task of the metaphysician, if she has any task at all, is tosystematically investigate the ways in which relatively separated and special tracks of 

scientific inquiry ‘hang together’ to imply a whole greater than the sum of their respective

 parts. This is important not just because people like having unified world-views. Principled,if always necessarily tentative, answers to metaphysical questions are required to help

scientists make sensible bets on which special-science kinds they should be trying to

explain and which ones they would be better advised to try to explain away.(justificar)

Study of unification as a distinctive enterprise can be predicted to co-vary in importancewith the extent to which individual sciences develop specially. In the heyday of positivism,

the demand for unification was typically given the strongest possible reading by

 philosophers who supposed that special-science generalizations should be logicallyderivable from more fundamental generalizations, and/or that all special-science types

should be logically constructible from fundamental types and relations. Insisting on such

‘strong unification’ amounts to asserting reductionism as a metaphysical hypothesis, which is just what, in disagreeing with Kim, we are here rejecting. As indicated above, the

history and practice of actual sciences, especially the behavioral, cognitive and life

sciences, honors no such reductionist constraint. However, the history and practice of 

science does demonstrate consistent concern for unification in a weaker sense. To theextent that the conclusions of a given special science are isolated from those of all other 

special sciences, in the sense that their  generalizations and/or ontological typologies are

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 65/100

strictly ‘brute facts’ from all available exogenous perspectives, we find ourselves with a

mystery or set of mysteries (Friedman 1974); and science is never content with mysteries.

We can try to describe the generic ambition for unification a bit more precisely bydistinguishing three specific kinds of project that might collectively constitute it:

1. Identifying a unifying ontological structure that justifies the argument patterns accepted

across all of the sciences.

2. Saying something genuinely enlightening about the ontologies of all sciences by

reference to general structural relations of some kind.

3. Identifying the ‘glue’ that holds all objective relations in place.

Notice that none of the three metaphysical problems we have identified is necessarily

about causation although all can be read as having something to do with it. For the time

 being (although see 4.3 below) to remain agnostic about whether the ‘glue’ might be

something worth calling causation. In recent philosophy of science the first problem has been most strikingly associated with the work of Philip Kitcher, the latter two with that of 

Wesley Salmon. We discuss relevant details of their respective positions shortly.

Talk of ‘binding ontologies together,’ or of the metaphysician’s ‘universal glue’ isunabashedly metaphorical. Positivism was, among other things, an attempt to explicate

unification without resort to superficial metaphor, but like most similarly motivated

 projects in the history of philosophy, it failed because it committed itself to claims that

were too strong and specific to fit the full complexity of actual science. We won’t, then, be

able to avoid metaphor here - `glue,’ indeed! - in trying to say what metaphysicalexplanation aims at. What we can do, and will, is as far as possible(en lo possible) allow

the dominant analyses in recent philosophy of science (Salmon’s and Kitcher’s) toconstrain what it is that we do with our metaphor.

4.2 Supervenience and Physical Causation

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 66/100

We saw in (3.2) how the prospect of a breakdown of ubiquitous one-way supervenience

struck Kim as tantamount to dualism. That shouldn’t be surprising, since we also saw in(2.1) that the first horn of the dilemma forming the supervenience argument has it that

denying supervenience just is denying the causal closure of physics. Kim demandssomething stronger than general commitment to supervenience in the form of a principlerequiring that there be ‘no changes without physical change’, though. He wants (see 1.1)

‘narrow’ supervenience, where the supervening properties of some entity must supervene

on its internal, or intrinsic properties. Failures of  this kind of supervenience don’t by

themselves imply anything about whether physics is causally closed. One way such

failures can arise, consistently with the closure principle, is from cross-classifying

taxonomies, which in turn can arise from triangulational individuation of mental

states, as discussed in (3.1) above.

Though triangulational individuation is compatible with cross-classification, it doesn’t

imply it. Social-ecological properties relevant to mental state individuation in the case of 

the pelorus operator (section 3.1) could be micro-based in Kim’s sense. Perhaps cognitivescientists could work adequately with a system of mental-state classification sensitive to

two or three micro-based taxonomies of properties among which it banned conflicts with its

own coherence rules.[xviii] However, cognitive scientists are just not, as a matter of fact,

trying to regiment their macro-properties in the way relevant to this scheme. (As Wallace

(2003) argues, physicists faced with the logically identical issue in relating quantum-level properties to macro-properties don’t try this either.) So this apparent possibility for 

reconciling Kim with cognitive science is worth pursuing only if Kim’s independent

metaphysical motivations for needing some such reconciliation are truly pressing. As wenow argue, they aren’t.

Kim provides no direct analysis of the concept of a physical property. Instead, as we

have seen, he assumes that the domain of physical properties is antecedently clear, and then

analyzes putative non-physical properties as micro-based macroproperties. Cognitive and behavioral scientists might imagine that this way of proceeding reflects consensus among

metaphysicians, appealing to some well-established analysis of the physical. This is not the

case. Recall, first, that the distinction between ‘top-down’ and ‘bottom-up’ prevalent in philosophy of science is drawn – by Kitcher and Salmon – by reference to the logic of 

explanation, not by appeal to a brute concept of the ‘physical’. We need to ask, then, what

makes something a macro-state, relative to some other set of states that are micro-states. Itwould be circular in this context to say that  M is a macro-state relative to micro-states m1,

… mk  just in case M is specified in terms of properties that supervene on the properties in

terms of which m1, …, mk  are specified; and Kim, given his project of  showing that

supervenience does not explain the relationship between the mental and the physical,would have to agree. Since we want to test Kim’s picture against the prevailing

metaphysic in general philosophy of science, we need to relate the macro-micro

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 67/100

distinction directly to differences in kinds of explanations. This can be done following

Kim’s own lead, as given in his r emarks on the relationship between explanation andcausation quoted in (3.1) above. Let us say that M is a macro-state relative to m1, …, mk  just

in case the (mere) information that  M  obtains fails to carry information picking out a

 particular member of m1, …, mk as causally relevant. Let us add that  M  is a  scientifically 

reputable macro-state just in case restrictions on the set of micro-states, one or more of  which must have obtained, can be stated in a scientific vocabulary more general than that of 

the special science that generalizes over states of type  M . The point of this way of 

restricting the scientifically reputable macro-states is to reflect the weak unificationrequirement that special sciences cannot be metaphysically comfortable in complete

isolation. Thus: an individual’s performing an action that constitutes a move in a social

game (e.g., the perolus operator’s uttering ‘[Landmark x] 237’) carries the information thatsome set of dispositions selecting the relevant action, encoded by the potentials along some

synaptic pathways in that individual’s brain, was available to be triggered, and was in fact

triggered, by some state of affairs encoded as an instance defined according to the rules of 

the social interaction by some other set of synaptic pathways in that same individual’s brain. By virtue of what might knowledge of the social action carry information about such

generic sorts of brain processes? By virtue of the actual and particular content of some

empirical theory of mental architecture and its relations to neural structures, on the onehand, and behavior, on the other. Similarly, to pick up the final example from (3.3), to say

that ‘hungry things are more likely to eat’ compresses information about a range of 

multiply realizable states and mechanisms, and is arguably not further compressible.

 Notice that this way of analyzing the macro-micro relation is strictly relative to a particular 

special-science context; we have said nothing yet about what might make some state or 

 property ‘intrinsically’ or ‘absolutely’ micro. This is because commitment to the non-

isolation of special sciences does not imply commitment to the idea that all special sciencesadmit of hierarchical analysis in terms of one basic science. Kim, however, must suppose

that there are `intrinsically’ micro states, since only this could justify his implicit restriction

on scientifically reputable macro-states being, as it is, stronger than the one just given. The point is not that he must suppose that psychological states reduce directly to such states;

rather, the claim is that if there are no such states to end a potential regress and do the ‘real’

causal work, then Kim’s supervenience argument would lead to an antinomy rather than toa disjunction with a preferred horn as he supposes. Our key question, then, is: does

(serious) metaphysics lend support to this intuition? Non-philosophers might be

disappointed, although not surprised, to hear that the answer is complicated. It requiresexamining some details of the tensions and complementarities in the two generic

 perspectives, the Kitcherian and the Salmonian, on the scientific realist’s epistemology.

Kim and his supporters can, prima facie, draw strong support from Salmon (1984, 1999)

whose most general goal is to articulate and defend a realist interpretation of the point andnature of science. In particular, according to Salmon, science aims to describe the causal

structure of the world. In the end, we’ll raise grounds for doubting that ‘causal’ is an

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 68/100

unambiguously perspicacious word here. So let us say for now that the essence of this sort

of realism seems to us to be crucial to any sort of realism worth having, and describe thatessence thus: Science aims to tell us how the world is structured, that is, how its

various processes and classes of entities constitute a single working machine .[xix] In

trying to describe how a machine works, a natural approach is to try to lay out its various

internal processes and indicate how they influence each other. Salmon aims to justify a picture of science that, as a whole, is engaged in this project. It is a virtue of such ambitiousrealism that it must go beyond mere affirmation of an independently existing world and

wrestle seriously with Hume’s epistemological challenge, to whit: How could anyone

know, by any amount of observation, which links between processes are causal and which

are not? Salmon’s answer here is that we can observe something that is preciselydiagnostic of causation. That is, we can see that certain processes transmit information

about their antecedent stages while others do not. Only the former are genuine processes.

Following Reichenbach (1957), we can put this in terms of the transmission of marks. Inthe absence of specific structure-preserving (and, ultimately, structure-constituting)

activity, entropy will eliminate marks on objects that carry information about their histories.

A structure is, by definition, something that resists entropy, even briefly. Therefore,wherever marks are preserved we have structure. The goal of science is to discover the

structures in nature. We can discover such structures because, as fairly sophisticated

information-transducing and processing systems, we can detect, record and systematicallymeasure mark-transmitting processes.

This is a terrifically powerful and, we think, deeply inspiring idea. It captures the core

component of  scientific realism – that science describes mind-independent natural

structure (and activity), in an ontologically systematic way – while respecting the essenceof empiricism. This latter constraint is that science has no place for inherently hypothetical

events or processes that are in principle beyond our capacity to physically detect, e.g.,

events on the other sides of space-like or time-like singularities, such as the interiors of  black holes or the far side of the big bang, or events outside of our collective light-cone.

One of us (Ross 2000) has exploited this idea to suggest a general metaphysic of existence;

so we could hardly think it more important as a metaphysical insight. What, though, does ithave to do with causation?

4.3 Two notions of causation

Salmon takes the idea described in the preceding section to be, first and foremost, an

analysis of causation. Is it? As our remarks immediately above make clear, it is certainlyan analysis of something[xx]  quite fundamental. But its primitive notion is information- 

transmission (in the physical and mathematical, not pragmatic, sense of ‘information’), not

causation. It therefore amounts to a semantic proposal to treat causation as an information-

theoretic concept. Should we accept the proposal? Since Salmon recognizes Hume’s

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 69/100

challenge to the effect that causation cannot be picked out by some observational procedure

independent of the analysis itself , this evaluation must proceed pragmatically. What effectwould accepting the semantic proposal have on our broader conception of science, and of 

 particular sciences? In particular, will it justify Kim’s intuitions about intrinsically micro-

causal relations?

Kitcher (1989) provides a detailed critique of Salmon’s analysis, which we willsummarize. First, we must reproduce Kitcher’s gloss of Salmon’s analysis:

(CP) P is a causal process iff there are spacetime points c, e such that P links c and e and it

is possible that there should be a modification of  P (modifying a characteristic that wouldotherwise have remained uniform) produced at c by means of a single local interaction and

that the modified characteristic should occur at all subsequent points from c to e withoutany subsequent interaction (1989: 462).

This rests the idea of a causal process on the prior idea of a causal interaction, demandingan analysis of causal interactions in non-causal terms. Here is Salmon’s analysis of causal

interaction:

(CI) Let P 1 and P 2 be two processes that intersect with one another at the spacetime point S ,which belongs to the histories of both. Let Q be a characteristic that process  P 1 would

exhibit throughout an interval (which includes subintervals on both sides of S in the history

of  P 1) if the intersection with  P 2 did not occur; let  R be a characteristic that process  P 2would exhibit throughout an interval (which includes subintervals on both sides of S in the

history of  P 2) if the intersection with P 1 did not occur. Then the intersection of  P 1 and P 2 at

S constitutes a causal interaction if: 

(1)  P 1 exhibits the characteristic Q before S , but it exhibits a modified characteristic Q’throughout an interval immediately following S ; and

(2)  P 2 exhibits the characteristic  R before S , but it exhibits a modified characteristic  R’

throughout an interval immediately following S . (Salmon 1984: 171).

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 70/100

We then have a case of causal interaction between  P 1 and P 2 at t iff there exist S , Q, R

at t satisfying CI. Two features of this analysis are crucial to Kitcher’s criticism. First, itdepends essentially on counterfactuals: we need to be able to pick out characteristics that

would   have carried on inertially in the absence of the interaction. Second, it makes the

concept of a macro-level cause depend on the idea of a micro-level cause. (This is just what

Kim assumes is unproblematic.)

These two features interact to generate the main criticism.[xxi] First, note that macro-

 processes typically involve vast ensembles of interactions. To use Kitcher’s example, if a

 batted baseball breaks a window, then we have, along with the interaction between the bat

and the ball, interactions between the ball and gusts of wind, the ball and changes in the

Moon’s gravitational field, these changes and the window, etc.. We thus need to be able to pick out the relevant counterfactuals to identify the macro-cause , viz.,

(A) If the bat had not intersected P 1 [the process that is the history of the ball’s spacetime

coordinates] then the momentum of  P 1 would have been different;

(B) If the momentum of  P 1 after its intersection with the bat had been different then the

momentum of  P 1 just prior to its intersection with  P 2 (the window) would have been

different;

(C) If the momentum of  P 1 just prior to its intersection with  P 2 had been different, then themomentum of  P 1 just after the intersection would have been different (specifically, the

window would not have broken!) (Kitcher 1989: 471).

But how do we know to pick out these counterfactuals ? By reference, it would seem, towhat we already know about the general causal structure of the world! Notice also that

if we have these counterfactuals picked out, then we might be tempted to analyze the causal

 process just in terms of  them; a detour through informational considerations would seemredundant.

A defender of Salmon could reply here that his analysis takes as its proper object only an

ideal  causal process, which would be a micro-process such that, given the restricted

predicates available in its physical description, S   is exhaustively and exclusively defined  

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 71/100

 by a finite set of characteristics Q, …,  Q’ and  R, …,  R’. Now, however, it seems that we

must know the causal structure of the world in order to pick out the class of idealinteractions S .

We do not know if these technical problems can ultimately be solved. For our purposes

here, this matters less than the general complaint Kitcher draws on the basis of them. That

is: Salmon’s analysis “requires that we provide an account of the way in which the causalstructure of the macroscopic world results from the stringing together of elementary

 processes. Even if we already had such an account, the emerging picture of our causal

knowledge is one in which the justification of recherché theoretical claims about idealized processes seems to be fundamental and our ordinary causal knowledge derivative” (Kitcher 

1989: 469). Now, we do not think that this constitutes a serious objection to Salmon’s

substantive accomplishment, where that is interpreted as articulating the kinds of realstructures in the world that science aims, in the limit, to discover. However, we do think 

that Kitcher’s point has force against the idea that an analysis such as Salmon’s, even if itcan be made technically bullet-proof with respect to its intended sphere of application in

fundamental metaphysics, can be pressed into service as an analysis of the elaboratelymacroscopic, feedback-driven processes cognitive and behavioral scientists seek to

characterize when they talk about mental causation, and the similarly complex causal

 patterns characteristic of science in general.

We will now present an alternative interpretation of Salmon’s achievement, intended to

shed light on what we see as the equivocal nature of the concept of causation. We take

our cue here from Redhead’s (1990, drawing on Kuhn 1971 and Russell 1917) discussionof causation and physics. Redhead notes that classical physics, in which forces played a

crucial role, has given way to forms of physical theory in which forces have been

eliminated. Redhead, we think justly, accuses those metaphysicians who wish to retain

forces of anachronistically clinging to a distinction between natural and forced motion. Ingeneral relativity, says Redhead:

There is no such thing as a non-natural motion. To most physicists the old-fashioned idea

of cause arises from the idea of our interfering in the natural course of events, pushing and pulling objects to make them move and so on. In modern physics there are just regularities

of one sort or another (Redhead 1990: 147).

This attitude represents a principle that seems to us to be well justified by induction on thehistory of science. The central concepts of  traditional metaphysics, including the

Aristotelean distinction between natural and forced motion, are  folk  concepts .

`Eliminativism’, in the usual sense of that word in the philosophy of mind, is the thesis that

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 72/100

folk concepts tend to be progressively eliminated from scientific practice. We don’t endorse

eliminativism in that sense . The concept of mind, for example, may enable us to pick 

out and generalize over real patterns in nature. Furthermore, the folk concept of agent

causation may be biologically necessary in the sense that no functioning agent could get by

(arreglarselas) without it. Science will therefore have things to tell us about both minds and

agent causation. However, there is no compelling reason to think that folk  intuitionsabout which patterns, if any, must be general should survive as the scope of scientific

knowledge widens. The concepts and axioms of Euclidean geometry pick out and organize

some real patterns – the class of (approximate) physical Isosceles triangles, for example.But Euclidean geometry is not general, in the sense of describing most space adequately.

We may gloss Redhead as suggesting that the concept of causation has its uses in

describing the doings of agents, and perhaps in a range of other special inquiries, but thatthese uses do not generalize to physics.

Redhead is clearly asserting that metaphysicians  should not  use the concept of causation in talking about general physical relations. This is a stronger conclusion than

we endorse. Suppose that a scheme developed from Salmon’s proves empirically adequateand logically perspicacious for bringing us closer to an analysis of the universal glue that

metaphysicians seek. Suppose furthermore that Salmon’s use of the term ‘causal structure’

to describe what he is analyzing sticks, and not simply out of semantic inertia, but because – after all – the idea of  analyzing causation as at bottom an informational relation isn’t

 silly, tonta,  or  pointless., inutil,  Then it would be right to say that the concept of 

causation had generalized. However, it would have done so along only one or a few of thedimensions that compose its historical semantic vector. Other such dimensions, those

 peculiar to the concept’s origins in describing the interventions of agents, would have been

discarded. Alternatively, we might end up (on a similar outcome in the philosophy of 

science) with ‘causation1’ and ‘causation2’. Our argument will not require a preferenceamong these or other semantically plausible scenarios. The claim we need is merely a bit of 

conceptual history: that causation has its origins as a folk concept associated with

agency, and that the concept as it figures in realist fundamental metaphysics, such asSalmon’s, is intended to have no such associations, since it must shed them if it is to do the

work Salmon wants from it. We will then see that Kim’s challenge to functionalism

depends on these very associations.

Before going further, it will help briefly to substantiate this conceptual history. Prior to the

modern period we find no concept equivalent or isomorphic to the kind of causal notion

analyzed by Salmon.[xxii]  Aristotle’s efficient causation, considered apart from his wider 

metaphysic, maps best onto Redhead’s ‘pushing and pulling’ of objects by agents. The full

Aristotelean story, with its multiply composed causes and deep teleology, is an elaborationof the folk notion modeled on the execution of a plan for intervention by an agent. The rise

of science disturbed this picture. Famously, the rationalists were led to continual

controversy amongst themselves over how to relate agent-causation to mechanicalaccounts; thus we have Descartes’s immaterial will that nevertheless exerts mechanical

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 73/100

effects, Malabranche’s occasionalism, Leibniz’s pre-established harmony, and so on. We

suggest that Hume’s attempt to analyze causation away is best interpreted in light of thishistory. Following Hobbes, but with much greater sophistication, Hume sought to explain

all mental activity as mechanical (Ross 1991). Furthermore, on his account all mental

activity had its ultimate impetus outside the mind, in the sources of impressions. Hume’s

was thus the most thoroughgoing, riguroso, denial of Aristotelean agency observed inWestern philosophy to that point. But the elaborated folk  notion of causation he

inherited from his tradition was rooted in the idea of the intrinsically active agent.

Attempting to drain this element out of the concept of causation, Hume found almostnothing positive left in it; and so, in his hands, it becomes merely a superstitious over-

interpretation of regularity.

Hume thus set the philosophy of science along a trajectory that, with respect to itstreatment of causation, finds maturity in the analyses of Reichenbach and Salmon. On this

whiggish reading of the history, Kant represents a regressive step, trying to regiment thefolk notion within the necessary operations of the understanding, and positivism a recovery

of the Humean path from within the framework of Kantian metaphysics (Friedman 1999).Had this been the only major development in theories of causation after Hume, then it

would be appropriate to describe the modern history of causation as a steady re-analysis

away from the original folk notion and towards an idea – whatever its exact content - thatcould find its conceptual gravity wholly within the framework of fundamental physics. In

evaluating efforts like Salmon’s as analyses of causation, we would then be asking, in

effect, whether the concept ultimately finds a role within that framework, or is fated for elimination.

However, this post-Humean development is not all that has happened to causation in recent

 philosophy. With the rejection of positivist and behaviorist accounts of mind in the 1960s

and 1970s, functionalists reasserted the metaphysical significance of the mental, in a waythat was not a re-working of Kant’s attempted compromise with empiricism. Functionalism

 – when it does not drift toward epiphenomenalism – seeks to give the mental a real and

distinctive causal role. Most importantly for present purposes, it understands that role in away that resurrects the folk idea of causation, since the minds defended by

functionalists are analyzed precisely as the ontological basis for agency. Thus, while

one tradition within the philosophy of science continued the project of trying to drain theagency out of causation, a parallel department worked assiduously at putting it back in! The

contemporary metaphysical muddles (desoreden, desparpajo) represented by Kim, against

which we are taking issue, are consequences of this double development.

As discussed in section (1), functionalists have disagreed significantly amongst themselvesover how mental causation could best be rehabilitated. The attributionist school of thought,

following Dennett, has articulated a metaphysics of mind according to which the

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 74/100

componential analysands of mind have been progressively distanced from the microcausal

gears of behavior – brains and nervous systems. This is the perspective whoseconsequences for behavioral and cognitive science is discussed in section (3). Philosophers

in this camp can excuse themselves from any particular commitments with respect to the

fundamental metaphysics of causation-in-general, except  insofar as some particular 

account of such causation turns out to be essential for respecting weak unificationconstraints on all sciences. Describing as they are an unabashedly (desenvuelto)

macroscopic set of phenomena, and with no ambitions in the direction of reductionism, [xxiii] 

they can respond to demands for explanation of mental causation in the same way as any 

special scientist asked to discuss the kinds of processes picked out by the scope constraints

of her discipline. Ask a geophysicist about `geological causation’ and you will be toldabout tectonic plates and flows of undersea lava and so forth. Similarly, a functionalist

cognitive scientist might address issues related to mental causation by talking about

feedback mechanisms, servosystematic control architectures, modules built by naturalselection, neural networks simulating Von Neumann computers, and so forth. To a

philosopher who regards the topic of mental causation as essentially a part of  

fundamental metaphysics, such answers will look like cases of changing, cambiante,vestuario, the subject. However, they are no more illegitimate than the geophysicist’s

similar answer to the similar question. The point here is just that a scientist’s ‘taking

metaphysics seriously’ does not imply slavery to semantic legislation by metaphysicians.Saying that special sciences are sensitive to metaphysical issues isn’t to say that special

sciences are exercises in (highly specific) metaphysical inquiry. If, in the end of the day,

metaphysicians convinced us that the concept of ‘causation’ descended from Hume is

more confusing than helpful, then either  cognitive scientists will find other ways to talk about evolved macroscopic patterns in  behavioral control built by interactions of genetic 

and cultural evolution, or, alternatively, we’ll collectively ‘decide’ to let the Aristotelean

semantic heritage triumph, regard cognitive scientists as having provided a naturalistic analysis of agent causation, and conclude that ‘causation’ is a concept restricted   to

application in cognitive science and other disciplines that study agents. From the present

state of play, it seems to us, either future semantic trajectory is possible; but neither threatens functionalism.

To return to the general point and summarize it, the history of philosophy incorporates a

tension between two quite different notions of causation, both of which survive because

 both have been intended to serve legitimate but differing projects. On the one hand, special

sciences are partly constituted by parochial types of ‘interaction-transmission’ relations,

where by such relations we refer to Salmon’s ‘glue’ without pre-judging the details of the

relationship between this and any causal concept as used in any particular special science.As Kitcher has emphasized, parochial, special science-relative varieties of the

interaction-transmission relations are reciprocal functions of accepted explanatory

schemata in the relevant sciences. Aristotelean agent causation, or folk psychological

causation, is one such special interaction-transmission relation. Contemporary

functionalism has significantly revised the Aristotelean or folk notion, particularly in

denying the coherence of the idea of a unified ‘Cartesian’ will with direct causal capacities

of its own, but there is a clear lineage relationship nevertheless. Those Kim charges with

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 75/100

 being ‘free-lunchers’ stop here, content to point out how useful this notion is. Kitcher’s

analysis of the importance of unification is one aspect of paying for lunch, in that it aims

simultaneously at explaining and transcending (without abandoning), the cluster of 

parochial interaction-transmission relations. That is, Kitcher takes the existence of this

cluster as a fact for metaphysics to explain. This is why Salmon, who was engaged directly

in constructive metaphysics, can view his project and Kitcher’s as complements.

Salmon’s project, however, simultaneously continues to appeal to the other philosophical

tradition with respect to the concept of causation and its use. As we explained above, the

origin of this tradition lies in Hume’s conviction that the folk concept of causation, asrooted in the kinds of program explanations peculiar to invocations of agent causation, fails

to generalize. That concept therefore fails to be a suitable candidate for universal glue.

Salmon is in pursuit of such glue, and this pursuit is the core of the serious metaphysician’s job. What potentially confuses matters is that Salmon thinks his candidate for glue

 preserves enough traditional associations to make ‘causation’ the appropriate name for it.We argued above that this is a semantic decision on his part. It is an understandable and not

unreasonable decision, but it is not forced science and it doesn’t commit a follower of Salmon to thinking that an analysis of some concept of causation deployed outside the

 project of seeking universal glue must be illegitimate. As we explained, that the substance

of Salmon’s effort can and should survive a revision of his semantic decision. Whether or not what Salmon gives us is best described as an analysis of `causation’, his work 

demonstrates the need for understanding the structure of the world in terms of objective

informational properties if one is to reconcile realism and empiricism . This involves noretreat from understanding Salmon’s work as a deeply illuminating contribution to

fundamental metaphysics; and the semantic revision need in no way obscure the fact that

the contribution evolves out of earlier inquiries, such as Hume’s and Reichenbach’s, into

the nature of causation. Ultimately, if Salmon’s work has shown us the way toward asuccessful account of  universal  glue, then all parochial special-science causal relations

must be susceptible to analysis in terms of it. This is what the definition of existence in

terms of information-transmission defended in Ross (2000), and cited back in Section (3),is supposed to achieve. Here, we do not depend on the unqualified success of that analysis.

Our ultimate goal – not yet accomplished – is to show that Kim’s whole project relies on

a metaphysical intuition about causation that is itself less secure than the role-

functionalist explanatory processes it seeks to undermine. We therefore need only

demonstrate alternatives to aspects of this intuition, not definitively to replace it with a

 better one.

However, the fact that we have had to reinterpret Salmon’s project in a special way mightstill seem worrying. All the weight of our answer to Kim, it may appear, rests on this

reinterpretation, so anyone finding it unpersuasive, or fearing that, whatever we choose to

call Salmon’s glue-candidate, it won’t be compatible with the cognitive scientist’s parochialconcept of causation, will be unsatisfied. Here we must work on locating the burden of 

argument. There is, first of all, no question that Salmon analyzes what he calls causation in

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 76/100

terms of information-transmission. So if Salmon-style analysis is to support Kim’s

reductionist picture, then we should be able to find an ‘information-transmission exclusion problem’ analogous to the causal exclusion problem. Defending such an analogy would

require a justified intuition to the effect that a map of all the real causal-transmission paths

in the universe has to have a simple, low-dimensional geometry, so that if information

 borne to a receiver by, say, the collisions of particles, were fully transduced and analyzed,then all information borne to that receiver down all other available paths would be

redundant. Can any of the work in fundamental metaphysics we have discussed here ground

such an intuition?

Perhaps. Suppose someone thinks that the sort of information-transmission relation

necessary for performing Salmon-style analysis of special-science causal relations just is 

the causal relation delivered by physics. In that case, the fact that science does observe arule to the effect that special sciences are not allowed to contradict the generalizations of 

 physics, conjoined with the view that the Salmon-style analysis in question isapproximately correct, would lead straight to the intuition just presented, and hence to

an information-transition exclusion problem. Kim indeed seems to believe that theserious metaphysician’s master-concept of causation, to which all special-science 

causation concepts are then answerable, comes from physics. This, at least, would explain

his convictions that micro- physical causes exclude mental ones, that supervenience must bea one-way relation, and that allowing some program explanations to count as causal

amounts to a form of dualism. A similar assumption underlies Jackson’s and Pettit’s

convolutions to the effect that program explanations may be ‘causally relevant’ but

cannot be causal . Our reinterpretation of  Salmon’s project allows it to go forward

unencumbered (no tener que cargar con algo) by this assumption that physics supplies one

concept of causation for every legitimate purpose. This is a good thing for that project,

 because the assumption is false of physics as we find it.

4.4 Physics and the physical

In common with much metaphysical philosophy of mind, Kim’s arguments trade on

(aprovecha) a particular image of how things are with physics. This image includescommitment to the view that there is no controversy about whether the apparently causal

claims of physicists are indeed causal,[xxiv] and that the distinction between physics and the,

‘special’, sciences is simple and exclusive. Much of the bite of the causal exclusion

 problem arises from the contrast between physics thus understood, and the special sciences.

Both assumptions, though, are false: physics in general is not enquiring into ultimate

causes, and much, perhaps all, of physics consists of a large collection of special

sciences.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 77/100

We do not deny that there is a metaphysically important sense in which physics is

fundamental. Physics is alone among the disciplines in being required to aim atgeneralizations that hold across all materially possible worlds (by which we mean: all

worlds that could actually exist within the boundaries of the singularities, space-like and

time-like, that limn in-principle observable spacetime). This leads to the fundamental

asymmetry in the structure of the sciences we mentioned above, according to which noother discipline may violate currently accepted generalizations of physics, but physics itself 

respects no corresponding limitation. This entails a (weak) version of the principle of the

causal closure of the physical: no special science may traffic in information-transmissionrelations, and, therefore (again, presuming the adequacy of a Salmon-style analysis), in

 parochial causal relations, that are spooky according to physics. But Kim, as we have

shown, needs something stronger than this. The intuition that brings Salmonesquemetaphysics into the service of his causal exclusion problem requires that physics  supplies 

the form of general causal relation that must then generalize. But it doesn’t.

Working physicists sometimes talk about causes. But, as Cartwright (1983, 1989) has

argued, this is because most working physicists, most of the time, are not in search of nomic generalizations holding across the whole scope of materially possible reality. They

are, therefore, working within special sciences within physics. If we ask, as our engagement

with Kim has now forced us to, ‘What is a physical cause, in general ?’, we must answer thequestion by reference to the part of physics that seeks generalizations good across the

whole scope of the science, that is, fundamental physical theory (general relativity and

quantum mechanics and electrodynamics). What turns up, by way of examples, is nothing.

In section (4.3) we referred to Redhead’s remarks on the elimination of forces from

 physics. Redhead goes further than that, and argues that much physics has very little to do

with causes. Instead of causal laws, he maintains, physicists are interested in finding

‘laws of functional dependence’ such as Boyle’s law, where pressure and volume co-existin certain specifiable ways without it making sense to say that one causes the other.

Galileo’s law describes the behavior of falling objects, but doesn’t identify any general

‘cause’ of the displacement of objects. Acceleration, for example, just defines the kinematicrelationship expressed by the law. A standard move is to say that the law measures a

‘force,’ and that that causes objects to fall. But, as Redhead (1990: 146) notes, since the

notion of ‘force’ derives directly from the Aristotelean analysis, this move adds no contentto Galileo’s law beyond transference of an anthropocentric metaphor. Redhead intends

skepticism about the idea of causation as a scientific concept altogether, but we need

not go this far for the sake of the argument here.  Nor need we claim that if some special

 branches of physics invoke parochial causal notions of their own, these must be equivalentto the Aristotelean folk notion – we claim that there is a plurality of special interaction-

transmission relations, not just a folk notion and a scientific one. What is important here

is the factual point Redhead makes about the practice of physics, which is that it doesn’tfeature the use of any general sort of thing – forces, fields, charges – that is a characteristic

kind of cause, picked out by physicists in contrast to other possible general kinds of causes.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 78/100

If this is plausible where classical physics is concerned, it is surely that much more

 persuasive when we attend to contemporary physics.

In fact, the message obtained from careful attention to physics is about as bad for Kim’shunch as can be imagined.  Loewer (2001: 323) joins us in noting that Kim requires a

“generation and production conception of causation,” but then writes:

The fundamental laws (for example, Schrödinger’s law) relate the totality of the physicalstate at one instant to the totality at later instants. The laws do not single out (señalan)parts

of states at different times as being causally related. If  S’  is the microphysical state in a

region R at time t’ and t’ is a time prior to t , then nothing less than the state S of the region R* that fills the backward light cone of  R can be said to produce S’ . We cannot say that one

event (one part of the physical state) produces another part since the laws do not connect parts in this way.

It gets still worse. Batterman (2000) argues that most theoretical  (as opposed to purelymanipulative) activity in physics consists in searching for physicists call `universalities’. By

this they do not mean, as a philosopher likely would, metaphysical principles necessarily

holding everywhere, but  physical  facts that allow them to extract “just those features of  systems, viewed macroscopically, which are stable under perturbations of their microscopic

details” (129). In particular, they search for suitably abstract topological characterizations

of systems in which basins of attraction emerge that corral microphysically heterogeneous processes around the universalities – for example, renormalization group fixed points

among Hamiltonians in Hamiltonian-space descriptions of fluids, gases, magnets,

 pendulums and other diverse systems that display `critical’ behavior with respect to phase-states. Thus, far from invoking generation-and-production causal relations at the micro

level to explain functional dependencies among physical states, physicists look for 

 principled (tener principios de, basado en)physical reasons for ignoring  most of  the aspects with which such relations might be identified.

A follower of Kim might object that this is all just epistemology. If physicists can extract

useful generalizations by ignoring lower-level causal detail, just like psychologists do, then

this is all to the good; but it doesn’t, and couldn’t, show that necessary causal work isn’tactually being done ‘down there’ where micro-events are generating and producing other 

micro-events. However, this interpretation is at best gratuitous, and at worst a contributor to

confused physics. Physicists do not  begin by identifying a micro-level of generation-and- production relations and then find bases for abstracting away from some of these relations.

They instead engage in measurement, manipulation and re-paramaterization of whole

systems until universalities emerge. Wallace (2003) argues that failure to  shake off 

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 79/100

(sacudirse) the intuition that ‘down there,’ under the level of system-level patterns that

show stability within restricted measurement time-scales, lies a realm where allmeasurement values are definite independently of scale, leads to apparent paradoxes and

 pseudo-problems. For example, there is a widespread belief that the established quantum

formalism is incompatible with definiteness of measurement at the macro-level – the

famous problem of Schröedinger’s cat - and so needs to be supplemented with empiricallyunmotivated parameters for finding connection principles that temporally link multiple

worlds, or link multiple observers into continuous minds, so as to allow for superpositions

of micro-states without corresponding macro-superpositions (a cat being simultaneouslydead and alive). Such mangling of the formalism to make it square with our old

metaphysical hunches has a severe cost in terms of physical theory: it “almost inevitably

spoils the relativistic covariance of the theory.” We best dissolve these insoluble dilemmas,Wallace suggests, by dropping the hunches and thinking of ‘reality’ as measurement-scale-

relative patterns in structural properties of quantum states ‘all the way down.’ (Reading

Wallace’s argument alone, one might worry that his point is itself just a qualitative

 philosophical hunch, but this is not so. Recent work by Nottale (1993, 2000), for example,gives formal details motivated from within physics.)

The above survey of physical ideas is not intended to represent a settled picture of or a

committed prediction about the metaphysical implications of contemporary physics. The point, rather, is this. Physics supplies no ‘master-concept’ of causation that is motivated

independently of some particular explanatory program . Physics does not encourage,

and may well even actively discourage – as Wallace effectively claims – the Salmonesqueinterpretation of  causation-as-metaphysical-glue on which Kim relies, unless that glue is

reinterpreted structurally . By `structurally’ we intend reference to networks of 

informational relations as suggested earlier, or to topological structures of fractal

spacetime following Nottale (1993, 2000), or to something else that features a key property incompatible with Kim’s hunches about physics, namely that basic

measurements are indexed by global rather than local analytic procedures, in the strictly

mathematical sense of these terms. (That is, measurement values are not indexed toneigborhoods of points.[xxv]) What all such interpretations have in common is that, far from

undergirding the one-way local supervenience that Kim transforms into reductionism as a

general metaphysical principle, they suggest it to be a scientific anachronism. There is

indeed a deep tradition of ‘causation as universal glue’ to which Kim implicitly appeals ;

but that tradition has found its most sophisticated contemporary expression in

analysis of  causation as a special type of information-transmission or other global-

structural relation.[xxvi]  One cannot derive a basis for insisting that physicalism implies

one-way supervenience from analysis of this concept. So Kim’s particular reductionist

image, from which follows all the trouble for special sciences identified in Section (3), endsup resting on a vague and unmotivated hunch about an unanalyzed class of absolutely

general causal relations.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 80/100

The upshot of our discussion to this point is negative: we have shown that there is no

rational basis for thinking that special sciences that traffic in multiply realized andmultiply supervenient functional kinds should expect to have to endure conceptual

revolutions under pressure for general unification of science, or because they presently fail

to track real or non-redundant causal relations, thereby missing the genuine explanations of 

the phenomena in their domains. We thus hope to have shown cognitive and behavioralscientists how, at least in general terms, to see off( se deshace, despide) the complaints of 

skeptical metaphysicians. However, we noted at the outset that a more straightforward way

of doing that, by a simple appeal to epistemic pragmatism,  has always been available tospecial scientists, and is no doubt the sociologically dominant response. The justification

for all the work we have been asking behavioral scientists to do in working through our

argument depends on our claim that metaphysics should  be taken seriously, that is,

can actually contribute something  positive  to cognitive and behavioral inquiry. The

 promises made at the beginning of the discussion, therefore, have not been discharged until,

now that we have swung (oscilar) a wrecking (destrozar) ball at the structures of 

conservative metaphysics of mind, we say something about how to build a scientificallyuseful structure above the rubble.(escombro)To this we therefore turn in our concluding

section.

5. Conclusion

The general consequences of the preceding discussion for behavioral and cognitive

scientists can be consolidated as consisting of one negative point and one positive one. In

order to state the latter from a clear basis, we begin here by summarizing the negativeupshot. Recall that we identified a general and important metaphysical task as that of 

identifying whatever it is that holds all objective relations in place, metaphorically calling

this a kind of glue. Recall also that we have distinguished two different senses of ‘cause’

discernible in the history of philosophical reflection on causation, and relevant in

different ways to metaphysics and reduction. Kim’s conviction that the special sciences

are answerable to physics amounts to the conviction that the particular  causal claims 

 produced by physics amount to statements about the metaphysical glue. He thinks that

physical causal claims are already metaphysically unimpeachable, and that that is the

reason why the causal claims of special sciences have to answer to them. But physics is

itself largely composed of special sciences, physicists aren’t best seen as in the business of discovering causes, and the primacy of physics does not consist in the fact that physicists

are, simply in virtue of being physicists, automatically doing fundamental metaphysics.

Kim is thus multiply mistaken.

The intuition among philosophers that ‘down there’, in physics, lies an unproblematic

and univocal concept of causation that can directly inform metaphysics runs very deep. As

we have seen, even Jackson and Pettit, whose work has been motivated by a concern to

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 81/100

defend and articulate the basis for the strong autonomy of special sciences, succumb to this

 picture when they assume – without argument or even much discussion –  that programexplanations, no matter how important they may be to scientific explanation, cannot be

causal. We have argued, however, that the metaphysical ground is nowhere close to

being sufficiently settled or uncontroversial to drive a set of conclusions as logically

strong, or as troubling for scientific practice, as the new reductionists imagine. Indeed, wesuggested in the last preceding section that current trends in physics and the philosophy of 

physics make their commitment to a localist  conception of ‘causal glue’ look like an

increasingly poor bet.

We thus claim to have shown cognitive and behavioral scientists how to see off 

(deshacerse) metaphysicians who are skeptical about the explanatory adequacy of their 

hypotheses and conclusions on grounds that these rely on ultimately reducible or eliminablecausal mechanisms. They can say: “We’re scientists, not metaphysicians. We aren’t trying

to explain in general and at one analytic stroke (estilo,trazo)how our macro-phenomenarelate to micro-phenomena. This doesn’t mean that we dismiss metaphysics as irrelevant;

we’d worry if you were right that we’re positing scientifically isolated mystery processes.But we don’t need to integrate ourselves with other sciences by identifying mental causes

with non-mental causes  –  there isn’t any single scientific concept of causation to govern 

this.  We’ll stay integrated by piecemeal (poco a poco) connections as we go, and if ametaphysician offers us a more general unifying principle that actually sheds potential light

on our subject –  mind and behavior –  then we’re all ears.” In light of our discussion’s

length and complexity, however, we can’t exactly claim that this gift (regalo,don) has comefor free. Many may be inclined to think that all we have done is provided a tediously

unnecessary justification for doing something they could always do, but without work:

ignore philosophers. A scientist who believes that all  metaphysics is gratuitous to her 

activity will not thank us for buying her a lunch she had no interest in eating.

We have operated from the assumption, however, that metaphysics can and should be taken

seriously as a part of, and for the sake of, (por algo) science . We thus owe some

demonstration of payoffs (ganacias) in these terms. There are, we think, two. First, cognitive and behavioral scientists do, like most special scientists outside of physics,

invoke and rely on distinctive causal concepts; but these are frequently implicit, and this

implicitness can and does complicate debates over investigative methods andinterpretations of conclusions. We think we are now in a position to say something

enlightening about the causal concepts at work in cognitive and behavioral science.

Second, special sciences, despite being separable by definition, do lean on (se apoyan,

 presión ejercen) one another in a variety of practically significant ways. We will be able tosay something useful about the details of that too.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 82/100

Virtually all models in the broad domain of  cognitive and behavioral science rest on the

idea that nervous systems, in interaction with environments, are engines that ‘produce’ behavior  and perhaps – a recurrently controversial point –  r epresentations. Behaviorists,

Gibsonians, some connectionists and many neuropsychologists have motivated their 

research programs by, and thus apparently made their importance hostage to, the conviction

that representations are the wrong sorts of things for carrying ultimate causal efficacy.Those who make more robust use of representational structures typically counter their 

skeptical  scientific critics by noting that computer programs manipulating representations

are undeniably causally significant, and that anti-representationalist projects are simplyrefusing to avail themselves of helpful resources. Still, it is usually conceded , the causal

sources of behavior can’t be representational ‘all the way down’; somewhere,

somehow, there must be a level of activity analogous to that of electrical circui ts incomputers that fully explains the ‘mental’ patterns. To suppose otherwise is to allow the

 possibility of dualism or magical emergentism or some similarly irresponsible license for 

seceding from the legitimating sphere of real science. Anyone who doubts that real

scientists worry about these issues, in the course of criticizing, defending and building newwork upon serious models, need merely review a sample of back issues of this journal.

The set of assumptions that drives these debates is distinguished from  Kim’s only in being

(typically) a bit less explicit. They arise because the legacy of two concepts of causation intension is a  general  inheritance , alive in psychology as well as philosophy. Functionalist

analyses of representations as making irreducible differences to behavior are piecemeal

vindications of the scientific significance of old-fashioned agent causation. This sort of causation is disturbingly (rompe el orden) unlike the kind of modern causation by

mechanical bumping (tope, taco, golpe) or (later) magnetic or gravitational pulling that

all of science is still often imagined to be ‘ultimately’ about. It seems to us that in much

of cognitive science explanation  by allowing mental control of physical behavior  isviewed as a kind of pragmatic compromise: ever so useful for getting work done, but

 someday …

Suppose, though, that the appropriate way of dissolving the tension is to allow a refined

and sophisticated kind of agent-causation as a parochial special-science type, while giving

up(renuncia) on the modern, generic, kind of causation. Such positive news for cognitive

scientists should be as surprising – if comforting rather than threatening – to some cognitivescientists as to Kim and his followers among philosophers. Yet so far as other sciences – 

emphatically including physics – and careful metaphysical speculation are concerned, it is

 just as plausible to suppose that the fundamental ontological structures governing all of 

science are global and structural as to suppose that they are local and mechanical.Contemporary cognitive and behavioral science is dominated by accounts of feedback-

driven servosystems and hypotheses about how natural and cultural selection can build and 

maintain them. It is very natural to suppose that such complex dynamics  must be ‘built outof’ simpler processes in an additive way. This, however, is a metaphysical assumption,

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 83/100

derived from meta-reflection on the history of science, which is not now standing up well

under concerted (concentrado, conjunta) pressure.

Here is an alternative metaphysical image: the dynamic patterns studied by the cognitiveand behavioral sciences are instantiations, at particular scales of metric identification and

measurement, of more global dynamics characteristic of the physical universe in general .

Recent work in mathematical physics, information theory and analytical metaphysics showshow to make this claim relatively precise and non-fuzzy from the perspective of 

mathematics,[xxvii] but it is more than a little boggling (incredulidad) with respect to

 prevailing intuitions about explanation. However, the current adventures of the

conservative metaphysicians can help to remind us that explanation by reference to

‘ultimate’ collisions of particles is no less logically puzzling ; we simply grew accustomedto it during the long march from the days of Galileo and Kepler. Let us be clear: we are not

here asserting  that, as matters have turned out metaphysically, the world is made of 

informational topologies (or some other kind of globally structuring manifold, mulitple)‘all the way down’ and demanding that everyone sign up. Scientists are justified in their

prevailing pragmatic intuition that metaphysical inquiry doesn’t generate clean,

resolute satisfactions of this sort, and that this is related to the grounds on which theyshould keep some distance from it in their pursuit of lasting (duradera) explanatory

accomplishment (logro,conclusion). However, metaphysical frameworks guide science as

constraints whether we like it or not. Furthermore, for reasons we will now briefly discuss

 by explicit reference to the causal concepts of cognitive science, some such constraints are,at any given time, necessary for scientific progress. It is equally crucial that these

constraints be allowed to evolve, to the extent of complete replacement over time. Such

constraint management is sufficiently delicate, and sufficiently important, to justify

philosophical activity. We will illustrate the point by reference to some actual recent

debates and activities in cognitive science.

We have already mentioned one way in which implicit localist metaphysics influencesactivity in cognitive science: it leads those who develop representationalist models to

constrain them by the idea that they must be amenable to vindication through

implementation in some set of ‘lower level’ local mechanisms . Often, all this amounts to

is that a few speculative paragraphs on possibilities for such implementation get tackedonto the backs of papers describing representationalist models; and this is hardly a problem,

if it is a problem at all, over which to get worked up. However, it expresses the fact that

most scientists do feel a responsibility not to leave their models isolated from the wider,unified explanatory project. Vague speculations about implementation are, at their limit,

lip-service acknowledgements of this. The principle becomes important to science when it

is taken truly seriously. The leading expression of genuine commitment to the principle thathas recurrently characterized work in behavioral and cognitive science is restriction of 

modeling approaches to domains of explanation that are taken to be already unproblematic

from the perspective of localist implementation.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 84/100

Glimcher (2003) has recently given us a history of neuroscience from this explicit

 perspective. The Sherrington program for explaining all ‘determinate’ behavior byreference to passive reflexes, each of which responds in isolation, according to fixed

condition-action rules, to a finite menu of possible stimulations, is as perfect an instanceof commitment to Kim-style localism as can be found anywhere in science. Since

Sherrington doubted, empirically, that all behavior is determinate in this way, he was a

dualist; note, then, that his dualism was not a directly metaphysical thesis, but a  scientific 

response given an implicit metaphysical constraint on hypothesizing. More interesting, as

Glimcher shows, is that grounds for doubt about the capacity of pure reflexology to explaineven the plausibly ‘determinate’ behavioral patterns were made evident (by Graham

Brown) during Sherrington’s lifetime, and more systematic critiques in the mid-twentieth

century, both theoretical and experimental, by von Holtz, Mittelstaedt, Weiss and Bernsteinaccumulated decisive refutation. Nevertheless, Glimcher argues, the most emulated and

 productive neurophysiological investigations right now - connectionist models of learning

using backpropagation, and Shadlen et al ’s (1996) celebrated work on visual perception of motion in monkeys are Glimcher’s examples – continue to honor Sherrington’s localist

 paradigm. The contemporary work of course invokes a range of new mechanisms

Sherrington could not have imagined; but the commitment to input-driven, localized, non-hierarchically governed processes remains in place.

Glimcher’s point is not that these studies don’t merit celebration or emulation. His point,

rather, is that neuroscientists, despite knowing that localism in their domain isn’t generally

true, and despite their not being willing to allow dualism , concentrate their best energieson such phenomena as can  best be modeled in localist terms . We now suggest that this

should be interpreted not as a retreat from unification with other special sciences but as an

indication of extremely serious commitment to it  given a prevailing, mostly implicit,localist metaphysic.[xxviii]

With Kitcher, Friedman, Kincaid and other philosophers we have cited approvingly in thecourse of this paper, we agree that special scientists are right to care about avoiding

completely isolated explanations. Here an overt (abierto) normative principle is in order: If 

what science mainly delivered (entregada, dispersa) was a chaos of scattered (dispersa.Esporadico) descriptions of unrelated phenomena, we would be justified in feeling

crushingly (aplastante) disappointed in it. This would be science as, at best, a pure under-

laborer to engineering, not a collective project for increasing our understanding of theuniverse. Like Glimcher, we intend no criticism of  cognitive scientists who respect this

norm by doing powerful work that preserves unity by leaving wider metaphysical

assumptions unchallenged. However, explanation is no virtue if we don’t care whether 

explanations are, in addition to being comprehensible, true. This implies that we mustn’t

leave metaphysical presuppositions unchallenged in practice unless philosophical

reflection convinces us that the presuppositions in question are actually justified. The

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 85/100

 justification of a metaphysical presupposition should rely mainly on its fruitfulness in

science, so commitment to localism was a healthy restriction for a long time. But in somesciences – in physics, in economics, in many parts of biology and cognitive science – that

time has passed. Recent experience and reflection suggests that explanation of local

 phenomena as instances of global dynamic structures is a viable alternative route to

unification. 

This is just what Kim and the conservative metaphysicians deny. Kim’s commitment to

fundamental metaphysics is expressed as the insistence that one does not solve the mind /

body problem  by offering  particular accounts of intentional processes in non-intentionalterms. Rather, (porsupuesto) one must try to explain how and why in general “mental

 properties and physical properties are related, and hopefully, ojala, also explain why they

are so related” (Kim 1998: 5). We agree. We also agree that functionalism cannot bevindicated as providing such an account by mere appeal to supervenience. But

functionalism could  license agent-causation as a legitimate, special-science-parochial,sort of causation after all, if more general accounts of informational or other topological

dynamics can show it to be non-mysterious – and the prospects here look promising (seeJuarrero 1999). Surely we have, up to an important standard of generality, explained how

mental properties and physically non-problematic properties are related if we produce

a broad account of feedback-driven servosystems and the ways in which evolution has

built nervous systems that support them. If dynamic systems theory is a way of doing

metaphysics – and that is what we are suggesting – then servosystematic control without

localist reduction is not isolated as a basis for explanation (and the same goes, in spades, for evolution). Here, perhaps, is the source of the technical tools through which the special

problem of mental causation and the general questions about universal glue find a

common logic of address. But the working problem of mental causation, as we see it, is the

very old problem of how agency is possible. ‘Causation’, in this context, meanssomething special: the processes, whatever they are, by means of which thoughts and

decisions, beliefs and desires, make a real difference in the world. We see no reason to

 believe that there is any more ‘general’ a way of addressing this problem than by theapproach of contemporary behavioral science, with its  plethora of servosystematic control

 processes grounded in neuroscience and ethology.

Kim’s problematic is what you get if you reify the folk and the post-Humean conceptsof  causation. On the one hand, you find yourself wanting (falta algo) to show how

interventions by agents can make a difference to what actually happens in the world. But

then, on the other hand, you insist that these interventions must be micro-processes, or 

decomposable into micro-processes, that agents must not just turn out to be programs.Well, we think it’s overwhelmingly likely that agents are programs , and they aren’t

anything else. Some mental states are reliable bearers of information about other mental

states, even though no particular state in the supervenience base of one is a reliable bearer of information about any other particular state in the supervenience base if the other. If 

something is a running ,funcionamiento, series of such states, then it’s a program,

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 86/100

something that acts and exists by compressing information. Indeed, living systems are only

 possible at all thanks to the fact that some of their states, including mental states in thosewith brains, extract and emit useful (accurate) information in compressed form. That

they can do this is empirically evident. It is also not, pace Kim, mysterious: thanks to the

dynamics of servosystematic feedback structures, multiple supervenience is possible (and

actual).

This contradicts nothing that physicists either presuppose or tell us. Physicists, like all

scientists, study patterns of compressed information at whatever scales they can be found,

not, by elision, omission, some non-existent level of `ultimate’ micro-banging (golpe9and colliding. (choque) This is good news, we take it, for most cognitive scientists. But the

full news is even better. When we ‘do’ metaphysics in the naturalist’s way – by standing

 back (retroceder, distanciar) and looking hard at collections of special sciences inabstraction – then moving our attention from the cognitive sciences to the physical ones

doesn’t involve a discontinuous leap, salto, from spooky ,spectral, or redundant causalrelations to good old-fashioned mechanical ones. We see instead convergent dynamical

accounts that can be swapped (cambiadob)across the boundaries of the many specialsciences in a profoundly interdependent intellectual market.

Acknowledgements:

Previous versions of this paper have been presented at conferences in Dubrovnik,

Stellenbosch and Ghent. We thank the audiences on those occasions. We are also gratefulto John Collier, Daniel Dennett, Harold Kincaid, James Ladyman, Ausonio Marras,

Veronica Ponce, and Alex Rosenberg for critical feedback and comments, to the editors and

referees of this journal for comments and criticisms of an earlier version of the paper, andto Nelleke Bak for careful reading of the text.

Notes

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 87/100

References

Achinstein, P. (1983) The Nature of Explanation, Oxford University Press.

Armstrong, D. (1981) The Nature of Mind and Other Essays, Cornell University Press.

Baker, L. R. (1993) Metaphysics and Mental Causation. In:  Mental Causation, ed. J. Heil

and A. Mele, Clarendon Press.

Batterman, R. (2000) Multiple Realizability and Universality.  British Journal for  Philosophy of Science 51: 115-145.

Bickle, J. (1998) Psychoneural Reduction: The New Wave, MIT Press / Bradford.

Birks, J. B. (1963) Rutherford at Manchester , W. A. Benjamin.

Block, N. (1980a) Introduction: What is Functionalism? In:  Readings in the Philosophy of  Psychology (volume 1), ed. N. Block, Methuen.

Block, N. (1980b) Troubles with Functionalism? In:  Readings in the Philosophy of  Psychology (volume 1), ed. N. Block, Methuen.

Block, N. and Fodor, J. (1972) What Psychological States are Not.  Philosophical Review8(2): 159-81.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 88/100

Brooks, R. A. (1991) Intelligence without Representation.  Artificial Intelligence 47: 141-

160.

Burge, T. (1993) Mind-Body Causation and Explanatory Practice. In:  Mental Causation,

eds. J. Heil and A. Mele, Clarendon Press.

Cartwright, N. (1983) How the Laws of Physics Lie, Oxford University Press.

Cartwright, N. (1989) Nature’s Capacities and their Measurement , Clarendon Press.

Cartwright, N. (1999) The Dappled World , Cambridge University Press.

Chalmers, D. (1996) The Conscious Mind , Oxford University Press.

Churchland, P. (1981) Eliminative Materialism and the Propositional Attitudes.  Journal of  Philosophy 78: 67-90.

Clapp, L. (2001) Disjunctive Properties: Multiple Realizations.  Journal of Philosophy, 98:

111 – 136.

Clark, A. (1997) Being There, MIT Press / Bradford.

Dennett, D. (1981) Three Kinds of Intentional Psychology. In:  Reduction, Time and 

 Reality, ed. R. Healey, Cambridge University Press. Reprinted in Dennett (1987).

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 89/100

Dennett, D. (1987) The Intentional Stance, MIT Press / Bradford.

Dennett, D. (1991a) Consciousness Explained , Little Brown.

Dennett, D. (1991b) Real Patterns. Journal of Philosophy, 88: 27-51.

Dennett, D. (2001a) Are We Explaining Consciousness Yet? Cognition 79: 221-237.

Dennett, D. (2001b) The Zombic Hunch: Extinction of an Intuition? In:  Philosophy at the New Millenium., ed. A. O'Hear, Cambridge University Press.

Dupré, J. (1993) The Disorder of Things. Harvard University Press.

Elder, C. (2001) Mental Causation versus Physical Causation: No Contest.  Philosophy and 

 Phenomenological Research 62 (1): 111-127.

Fodor, J. (1968) Psychological Explanation, Random House.

Fodor, J. (1974) Special Sciences, or the Disunity of Science as a Working Hypothesis.

Synthese 28: 77-115.

Fodor, J. (1975) The Language of Thought, Harvard University Press.

Fodor, J. (1987) Psychosemantics, Cambridge, Mass.: MIT Press.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 90/100

Fodor, J. (1994) The Elm and the Expert , MIT Press / Bradford.

Friedman, M. (1974) Explanation and Scientific Understanding.  Journal of Philosophy 71:

5-19.

Friedman, M. (1999) Reconsidering Logical Positivism, Cambridge University Press.

Garfinkel, A. (1981) Forms of Explanation, Yale University Press.

Gintis, H. (2000) Game Theory Evolving , Princeton University Press.

Glimcher, P. (2003) Decisions, Uncertainty, and the Brain, MIT Press / Bradford.

Hempel, C. (1965) The Logic of Scientific Explanation, Free Press.

Horgan, T. (1997) Kim on Mental Causation and Causal Exclusion.  Philosophical 

 Perspectives, 11: 165-184.

Hull, D. (1972) Reduction in Genetics – Biology or Philosophy?  Philosophy of Science, 39:

491-499.

Hutchins, E. (1995) Cognition in the Wild , MIT Press / Bradford.

Jackson, F., & Pettit, P. (1988) Functionalism and Broad Content. Mind 97: 381-400.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 91/100

Jackson, F., and Pettit, P. (1990) Program Explanation: A General Perspective.  Analysis 50:

107-117.

Juarrero, A. (1999) Dynamics in Action, MIT Press / Bradford.

Kim, J. (1993) Supervenience and Mind . Cambridge University Press.

Kim, J. (1998) Mind in a Physical World , MIT Press / Bradford.

Kincaid, H. (1997) Individualism and the Unity of Science, Rowman and Littlefield.

Kitcher, P. (1976) Explanation, Conjunction and Unification.  Journal of Philosophy 73:

207-212.

Kitcher, P. (1981) Explanatory Unification. Philosophy of Science 48: 507-531.

Kitcher, P. (1989) Explanatory Unification and the Causal Structure of the World. In:Scientific Explanation, eds. P. Kitcher and W. Salmon, University of Minnesota Press.

Kitcher, P., & Salmon, W., eds. (1989) Scientific Explanation,  University of Minnesota

Press.

Lewis, D. (1972) Psychophysical and Theoretical Identifications.  Australasian Journal of 

 Philosophy 50: 249-258.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 92/100

Lewis, D. (1980) Mad Pain and Martian Pain. In:  Readings in the Philosophy of 

 Psychology (volume 1), ed. N. Block, Methuen.

Loewer, B. (2001) Review of Kim:  Mind in a Physical World .  Journal of Philosophy 98:315-324

Marcus, E. (2001) Mental Causation: Unnaturalized but not Unnatural.  Philosophy and 

 Phenomenological Research 63 (1): 57-83.

Marras, A. (2000) Critical Notice of Kim:  Mind in a Physical World . Canadian Journal of 

 Philosophy 30:137-160.

Marras, A. (2002) Kim on Reduction. Erkenntnis 57(2): 231-257.

McClamrock, R. (1995) Existential Cognition. University of Chicago Press.

McGinn, C. (1991) The Problem of Consciousness. Oxford University Press.

Menzies, P. (1988) Against causal reductionism. Mind 98: 551-574.

Meyering, T. (2000) Physicalism and Downward Causation in Psychology and the SpecialSciences. Inquiry, 43:181-202.

Millero, F. J. (2001) The Physical Chemistry of Natural Waters, Wiley-Interscience.

 Nagel, E. (1961) The Structure of Science. Harcourt, Brace and World.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 93/100

 Needham, P. (forthcoming) The Discovery that Water is H2O. International Studies in the

 Philosophy of Science.

 Nottale, L. (1993)  Fractal Space-Time and Microphysics: Towards a Theory of Scale-

 Relativity, World Scientific.

 Nottale, L. (2000) Scale Relativity, Fractal Space-Time and Morphogenesis of Structures.In: Sciences of the Interface: Proceedings of International Symposium in Honor of O. Rössler, eds. H. Diebner, T. Druckrey and P. Weibel, Genista.

Oppenheim, P. and Putnam, H. (1958) Unity of Science as a Working Hypothesis. In:

 Minnesota Studies in the Philosophy of Science, vol. 2. eds. H. Feigl, G. Maxwell and M.Scriven, University of Minnesota Press.

Papineau, D. (1993) Philosophical Naturalism. Blackwell.

Pettit, P. (1993) The Common Mind . Oxford University Press.

Place, U. T. (1956) Is consciousness a brain process? British Journal of Psychology 47: 44-50.

Ponce, V. (MS) Chemical Kinds, Microscopic Essences, and Chemical Laws.

Putnam, H. (1963) Brains and Behavior. In:  Analytical Philosophy, Second Series, ed. R.Butler, Basil Blackwell & Mott.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 94/100

Putnam, H. (1967a) Psychological Predicates. In:  Art, Mind and Religion, eds. Captain, W.

H. and Merrill, D. D. University of Pittsburgh Press.

Putnam, H. (1967b) The Mental Life of some Machines. In:  Intentionality, Minds and  Perception, ed. Hector-Neri Castañeda, Wayne State University Press.

Putnam, H. (1975a) Philosophy and Our Mental Life. In:  Mind, Language and Reality:

 Philosophical Papers, vol.2. Cambridge University Press.

Putnam, H. (1975b) Mind, Language and Reality: Philosophical Papers, vol.2, Cambridge

University Press.

Pylyshyn, Z.W. (1984) Computation and Cognition: Towards a Foundation for Cognitive

Science, MIT Press.

Raynor, H. A. and Epstein, L. H. (2001) Dietary Variety, Energy Regulation, and Obesity.

 Psychological Bulletin 127 (3): 325-341.

Redhead, M. (1990) Explanation. In:  Explanation and Its Limits, ed. D. Knowles ,Cambridge University Press.

Reichenbach, H. (1957) The Philosophy of Space and Time. Dover.

Ross, D. (1991) Hume, Resemblance and the Foundations of Psychology.  History of 

 Philosophy Quarterly 8: 343-456.

Ross, D. (1997) Critical Notice of Ron McClamrock:  Existential Cognition. Canadian

 Journal of Philosophy 27: 271-284.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 95/100

Ross, D. (2000) Rainforest Realism: A Dennettian Theory of Existence. In:  Dennett’s

 Philosophy: A Comprehensive Assessment , eds. D. Ross, A. Brook and D. Thomposon ,MIT Press.

Ross, D. (2001) Dennettian Behavioural Explanations and the Roles of the Social Sciences.

In: Daniel Dennett , eds. A. Brook and D. Ross, Cambridge University Press.

Ross, D. (forthcoming) Chalmers's Naturalistic Dualism: A Case Study in the Irrelevance

of the Mind-Body Problem to the Scientific Study of Consciousness. In: The Mind as

Scientific Object, eds. C. Ernelling and D. Johnson, Oxford University Press.

Rowlands, M. (1999) The Body in Mind , Cambridge University Press.

Russell, B. (1917) On the Notion of Cause. In: Mysticism and Logic, Allen and Unwin.

Salmon, W. (1984) Scientific Explanation and the Causal Structure of the World , PrincetonUniversity Press.

Salmon, W. (1990) Scientific Explanation: Causation and  Unification. Critica Revista Hispanoamericana de Filosofia 22: 3-21.

Salmon, W. (1999) Causality and Explanation. Oxford University Press.

Shadlen, M., Britten, K., Newsome, W., & Movshen, J. (1996) A computational analysis of the relationship between neuronal and behavioural responses to visual motion.  Journal of 

 Neuroscience 16: 1486-1510.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 96/100

Smart, J. J. C. (1959) Sensations and Brain Processes. Philosophical Review 68, 141-156.

Sorabji, R. (1988) Matter, Space and Motion. Cornell University Press.

Spurrett, D. & Papineau, D. (1999) A note on the completeness of ‘physics’.  Analysis 59(1): 25-29.

Spurrett, D. (1999) The Completeness of Physics. Doctoral dissertation at the University of 

 Natal, Durban.

Spurrett, D. (2001a) Cartwright on Laws and Composition.  International Studies in the

 Philosophy of Science 15 (3): 253-268.

Spurrett, D. (2001b) What Physical Properties Are.  Pacific Philosophical Quarterly 82(2):

201-225.

Stich, S. (1983) From Folk Psychology to Cognitive Science. MIT Press / Bradford.

Thalos, M. (2002) The Reduction of Causal Processes. Synthese 131: 99-128.

van Brakel, J. (2000) The Nature of Chemical Substances. In: Of Minds and Molecules: New Philosophical Perspectives on Chemistry, eds. Nalini Bhushan and Stuart Rosenfeld,

Oxford University Press.

Van Fraassen, B. (1980) The Scientific Image, Oxford University Press.

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 97/100

Van Gulick, R. (1993) Who’s in Charge Here? And Who's Doing All the Work? In:  Mental 

Causation, ed. J. Heil and A. Mele, Clarendon Press.

Wallace, D. (2003) Everett and Structure. Studies in History and Philosophy of Science B:Studies in History and Philosophy of Modern Physics 34: 87-105.

Wilson, R. (1995) Cartesian Psychology and Physical Minds, Cambridge University Press.

Yablo, S. (1992) Mental Causation. Philosophical Review 101 (2): 245-280.

[i] The phrase is due to Bickle (1998). However, we will not here be

engaging with Bickle’s interesting thesis, which has enough directempirical content to be a piece of cognitive science in its own right. Thephilosopher who has done most to inspire the backwash is Jaegwon Kim,and his most influential argument, as given in Kim (1998) will be thetarget of our discussion.

[ii] Or, at least, so philosophers often say. As a claim about actualbehaviorist psychologists, this claim is largely nonsense, flatly untrue of,for example, E.C. Tolman or Karl Lashley. But the importance given toknocking over this straw man in the history of the rise of functionalism isindisputable, and is what is of relevance to us here. We would encouragemore footnotes like this one in the philosophical literature, however.

[iii]

Functionalism, thus understood, can be a kind of behaviorism - justone allowing for some intermediate behaviors between stimulus andresponse. Our own favored variety of functionalism is in fact of thisbehaviorist sort; but this will not play any direct role in our argument inthis paper.

[iv] We use this example because it is standard in the literature we aredescribing. We should point out, though, that the philosophers whointroduced it knew from the outset that it was at best neurologically

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 98/100

implausible, and were using it as a place-holder for some imaginedfuture reduction of a psychological to a neurological state.

[v] The same argument structure has also been (e.g. Horgan 1997)used to argue for functionalism within a species (on the basis of significant neural and other differences between conspecifics), or, over

time, within a single individual.[vi] There are various particular ways of being a realizer functionalist inthe broad sense indicated here. One, particularly strong, way is via the‘functional analysis’ strategy associated with Armstrong and Lewis, asdiscussed in section (2.2). Another way, canonically defended inPylyshyn (1984), requires that types be individuated either by referenceto intrinsic properties of members of the type, or by reference tointrinsic properties of independently specifiable sets of tokens of thetype.

[vii] On our broad conception of realizer functionalism; see note (6)above.

[viii]

Kim (1998) is not the first expression of the problem, merely anelegant, sophisticated and up to date version of it. Yablo (1992) is aclear and widely cited statement of the issues as of the early 1990s, andthe papers in Kim (1993) show many of the lines of argument andthinking that lead up to Kim (1998).

[ix] We assume throughout that we’re talking amongst people whowould regard the admission of supernatural causes into science as theend of the world.

[x] Talk of ‘enhancing’ is somewhat sloppy, as Marras (2002) pointsout, but the details need not detain us here.

[xi] The philosophical literature on explanation is enormous, and so

some philosophers might object to our announcing that we can boil itdown to consideration of just two approaches. A few meta-comments onthat literature are therefore in order. It divides naturally into two piles. The first pile, concerned directly with the way in which the search forexplanation descriptively and normatively guides scientific activity,really does mainly revolve around the dialectic established by Kitcher’sand Salmon’s long argument with each other (see Kitcher and Salmon1989). The second pile, highlights of which include van Fraassen (1980),Garfinkel (1981), and Achinstein (1983), concerns the logic of explanatory statements. Both piles descend from the classic work onexplanation in philosophy of science by Hempel (1965), which, in the

way of positivism, saw these two concerns as indistinguishable. To apost-positivist of whatever stripe, however, they are distinct, and to aconsiderable extent orthogonal. That is, just about any combination of views from the first and second debates can be made compatible. (See,for example, Kincaid 1997, Chapter 5, whose work depends on subtlerecombinations of them.) For our purposes in this paper, only the firstset of issues about explanation are directly relevant. Non-philosophers

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 99/100

are cautioned, however, against taking our summary as a mini-survey of the whole literature on the subject.

[xii] As Batterman (2000: 118, n. 4) notes, “Kim's argument won't gothrough (es aprobado) unless the causal properties of themacroproperties just are the resultant (or `sum') of the microstructural

properties.”[xiii] For a sample of the literature urging this perspective, seeMcClamrock (1995); Wilson (1995); Clark (1997); and, especiallyinfluential with respect to what we say here, Dennett (1991a). Pettit(1993) provides the most systematic, though very cautious,investigation of these ideas.

[xiv] According to Menzies (1988) this line of argument was suggested byLewis.

[xv] i.e. A property possessed by an object (such as dormativity) invirtue of its having some more basic (e.g. chemical) properties.

[xvi] We are especially indebted to Ponce’s treatment here.[xvii]

Clapp (2001) successfully argues that some leading defenses of theautonomy of special sciences, such as Fodor’s (1974), are guilty of thislapse of metaphysical seriousness. We should therefore note explicitlythat none of our arguments in this paper depend on the idea that thekinds of any special science must be preserved as kinds just becausepeople find it useful to think with the concepts they represent. Indeed,on one interpretation this is what `taking metaphysics seriously' in oursense here means.

[xviii]  Stich (1983) devoted a book to arguing that this sort of picture,intended as a way of reconciling a plausible cognitive scientific typologyof states with folk psychology, couldn’t work. Kim must believe that Stich

is wrong about this.[xix] Cartwright (1983, 1999) has famously argued that the world is not asingle, working machine, but is instead ‘dappled’, by which she meansontologically disunified. Dupré (1993) has urged a similar thesis. Forreasons given in Spurrett (1999, 2001a), we reject this conclusion. Thefact that science is never finished, and therefore never completelyunified, may mean that its current description of the world at any giventime will always be of a world that is ‘dappled’; but to derive as ametaphysical conclusion the claim that the world is dappled is to simplyabandon the regulative ideal that informs Salmon’s project, and, for thatmatter, Kim’s. Answering Kim this way would simply amount to

shrugging off the significance of realist metaphysics, another way of trying to have lunch for free.

[xx] The ‘something’, we would say, is indeed fundamental structure;Ross (2000) takes it to be the network of Schrödinger-style negentropicrelations. That network is our favorite candidate for universal glue.

[xxi] Kitcher develops, at length, additional criticisms based oncounterexamples to Salmon’s technical criteria for distinguishinggenuine causal processes from pseudo-processes. We will not

7/29/2019 A critique of Jaegwon Kim.doc

http://slidepdf.com/reader/full/a-critique-of-jaegwon-kimdoc 100/100

incorporate these into our summary here, since they contribute little tothe issues relevant to our discussion, and since even if Salmon’sapparatus is repaired so as to block the counterexamples, Kitcher’s maincritique is unaffected.

[xxii] We rely here – with liberal interpretive work of our own – on Sorabji

(1988).[xxiii] It is common outside philosophy for Dennett to be called a`reductionist’ because he analyzes intentionality and consciousnesswithout recourse to any entities or processes incompatible with thecausal closure of physics. However, Dennett in fact denies, like us, thatthere is any general relation between physics and special sciencesstronger than global supervenience; and in the context of most debatesin philosophy of science, this makes him as anti-reductionist as therecent tradition allows. Thus, for example, when Kincaid (1997, 86-90)defends anti-reductionism, he feels he needs to spend a few pagesshowing that he needn’t go as far  in that direction as the `radical’

Dennett. Ross (2000) explains in detail the sense in which Dennett’santi-reductionism is radical.[xxiv] Philosophers typically grant that our current physical theories are

open to revision, so the point here is slightly more complicated. Still, forphilosophers of mind an ideal physicist is generally assumed to bemaking unproblematically causal claims, whereas an ideal economist,say, would need to do additional philosophical work over and above hereconomics to justify thinking of her claims as causal.

[xxv] We owe this insight to Andrei Rodin.[xxvi] The logic of this, and comparison of the causation concept’s role in

different branches of science, is made formally explicit in a recent paper

by Thalos (2002).[xxvii] We allude to the earlier references to the work of Nottale (1993,2000).

[xxviii] This point has also been vividly argued by Dennett (1991a).