Beyond Reduction: From Naturalism to Cognitive Pluralism

c© 2014 Imprint Academic Mind & Matter Vol. 12(2), pp. 197–244

Beyond Reduction: From Naturalismto Cognitive Pluralism

Steven HorstDepartment of Philosophy

Wesleyan University, Middletown, USA

Abstract

One of the most popular and long-lived approaches to natu-ralizing the mind is the attempt to reduce mental phenomena tophysical, biological, or neural phenomena. Reductionism has helda special allure among philosophers for a number of reasons: theintuitive appeal of part-whole explanations, reductive explanation’sresemblance to the axiomatic method in mathematics, its apparentpromise as a strategy for unifying different knowledge domains, andthe fact that true reductions, when successful, are almost uniquein yielding both metaphysical necessity and complete explanations.Reductionism was a very influential view in philosophy of science,both in early modernity and through much of the twentieth cen-tury, and some central contemporary issues in philosophy of mind– the explanatory gap and the hard problem of consciousness – areframed as claims that conscious mental states (and perhaps theyalone) are not reducible to physical phenomena.

In fact, however, most philosophers of science today would agreethat true intertheoretic reductions are rare even in the natural sci-ences. I propose an explanation of both the appeal and the failureof reductionism in terms of a cognitivist approach to philosophy ofscience called “Cognitive Pluralism”, and then explore what impli-cations post-reductionist philosophy of science has for philosophyof mind. If it is “explanatory gaps all the way down”, what arethe implications for dualism and for reductive, non-reductive andeliminative physicalisms? Is the mind-matter gap different from theother explanatory gaps? And, if the Cognitive Pluralist analysis iscorrect, is there any hope more generally for a “unified science”, orthat scientific theories generally can provide answers to metaphys-ical questions?

1. Varieties of Naturalism: A Brief Overview

There are a variety of philosophical doctrines that have gone by thename of “naturalism”. Some usages – such as the use of “naturalism” asanother name for Aristotelianism – are now largely archaic. But there arecontemporary uses of the term in ethics, metaphysics, epistemology, phi-losophy of mind and philosophy of science, and the positions thus named

198 Horst

represent a surprising variety of philosophical views. The applications ofthe word, while diverse, are not arbitrary, and it is possible to bring someorder to the situation by making a small number of distinctions.

One distinction is between two importantly different uses of the word“naturalism” in contemporary philosophy. The first, which is the pre-dominant usage in epistemology and philosophy of science, has its rootsin Quine’s (1969) “Epistemology Naturalized”, and is characterized by therejection of the use of a priori philosophical principles to guide science.Quine (1992, p. 6) writes that “naturalism looks only to natural science,however fallible, for an account of what there is and of what what there isdoes.” Rosenberg (1996, p. 4), in a survey article on naturalism, expandsupon Quine’s theme:

We may characterize naturalism in philosophy as follows:

1. The repudiation of “first philosophy”. Epistemology is notto be treated as a propaedeutic to the acquisition of furtherknowledge.

2. Scientism. The sciences-from physics to psychology and evenoccasionally sociology, their methods and findings-are to bethe guide to epistemology and metaphysics. But the morewell-established the finding and method the greater the re-liance philosophy may place upon it. And physics embodiesthe most well-established methods and findings.

In addition to epistemologists and philosophers of science, there arephilosophers of mind and metaphysicians who would endorse this sort of“methodological” or empirical naturalism as well (the latter terminologyis from Horst 2009). But in those fields, there is also a second usage,typified by commitments to particular types of explanation and/or meta-physics. To characterize this type of naturalism in a fashion broad enoughto include the many different attempts to naturalize the mind, it is neces-sary to begin with a schematic first approximation and then explain howit can be filled out in different ways.

General Schema: Philosophical naturalism in philosophy ofmind is the view that all mental phenomena are to be accom-modated within the framework of nature as it is understoodby the natural sciences.

Of course, some of the great figures in the history of modern sci-ence, such as Descartes and Newton, were also committed to paradigmat-ically supernatural beings and phenomena: God, souls, free will, miracles.Descartes actually used theological premises to derive substantive scien-tific conclusions, such as conservation of motion and the impossibility ofa vacuum. And while Newton kept theological reasoning out of the mainbody of his Principia and Optics, it is clear from his correspondence and

Beyond Reduction 199

unpublished manuscripts that his view of “the framework of nature” wasfraught with the supernatural. Such views would need to be excludedfrom the naturalist camp, as an important additional theme of natural-ists since early modernity has been an opposition between naturalism andsupernaturalism. Philosophical naturalism is, among other things, anti-supernaturalism. I would also (and perhaps more controversially) excludefrom naturalism any sort of view, however committed to particular princi-ples of the sciences, that does not regard the fundamental entities positedby the sciences as ontological bedrock, but instead tries to cash them outin terms of ideas or practices – say, various forms of idealism, pragmatism,or social constructionism. As a result, it is necessary to add a caveat tothe General Schema:

Caveat: A view cannot count as philosophically naturalisticif it (a) allows the existence of paradigmatically non-naturalentities such as God, angels or Cartesian souls or non-naturalproperties, or (b) treats the world of nature as understoodwithin the sciences as non-fundamental.

The Schema, however, is (intentionally) ambiguous along at least threedimensions. We may note these by way of three further questions, theanswers to which help to distinguish several varieties of naturalism:

1. How are we to understand the phrase “the framework of nature asit is understood by the natural sciences”?

2. Is the claim about “accommodation” a claim about explanation ora claim about metaphysical determination, or both?

3. Is the general schema to be understood as a positive claim (that themind can be so accommodated) or as a normative claim ( that itmust be so accommodated, or else some dire consequences follow?

There are a variety of philosophical views about how “nature” is “un-derstood by the natural sciences.” Some emphasize the role of naturallaws in science. Others emphasize evolutionary explanation. Still othersemphasize, not specific scientific laws or theories, but “science” as a com-prehensive framework in which particular scientific theories are united byinter-theoretic reduction to a common base. It is reductive naturalismthat will be my main focus in this paper. I think that this focus is neithermisguided nor idiosyncratic, because reductive naturalism has historicallygarnered the lion’s share of philosophical attention. And the fact that ithas done so is no accident.

A view of science based in laws need not be wedded to a metaphysicalnaturalism. A Cartesian dualist can believe that there are laws govern-ing mind-body interaction, and Newton believed that God is continuallyworking miracles. And evolutionary explanation of a feature of the mind

200 Horst

is viable only to the extent that we assume that there are purely naturalmechanisms for its heritability and expression as a phenotypic trait, thevery sorts of things called into question when phenomena like conscious-ness are claimed to be irreducible. (Horst 1999) In addition, philosophershave traditionally been interested in claims that are not merely empiri-cally adequate, but metaphysically necessary and sufficient. Reduction,and arguably reduction alone, provides metaphysically sufficient natural-istic explanations.

Here we pass to the second point of clarification: the distinction be-tween claims about explanation and claims about metaphysics. I takeexplanation to be marked by a kind of a kind of cognitive achievement,whereas metaphysical claims are claims about the nature of the world.There are forms of explanation, such as statistical explanation, that donot involve metaphysical determination. And there may well be forms ofmetaphysical determination that do not have corresponding forms of ex-planation, and may indeed be inscrutable to human minds (McGinn 1983,Pinker 1997). Inter-theoretic reduction is a particularly strong form of ex-planation, and is, again, one that has robust metaphysical implications,which we might express in the following principle:

Positive Explanation-to-Metaphysics Connection (EMC):If A is reducible to B, then B → A is metaphysically necessary,and A is metaphysically supervenient upon B.

The principle is, of course, phrased as a conditional. But why believethat some particular phenomenon is reducible to something else? Or,more broadly, why believe that some particular phenomenon can be natu-ralized? There are two basic kinds of answers, involving what I shall referto as positive and normative claims. By a positive thesis I mean simply athesis about what is the case. A normative thesis, by contrast, is aboutwhat must be the case. Sometimes reductionism and other naturalismsare offered as positive theses – as second-order empirical claims about howthings have turned out or will turn out in the sciences, such as the claimthat thermodynamics is reducible to statistical mechanics.

But reductionism and other naturalisms are also sometimes offered ina normative voice. Early Modern philosophers held that a study of thenatural world must take the form of demonstration from first principlesto count as scientia. Many of the Logical Positivists believed that uni-fication by reduction was required by “the logic of science.” The debatebetween reductionists and eliminativists in philosophy of mind is pred-icated on a shared assumption that, if a theoretical posit of a specialscience is not reducible, it is not real and should be eliminated. I willnote in passing that these normative forms of naturalism would seem tobe based in precisely the sort of a priori theorizing that is rejected by the


methodological naturalism associated with Quine. Quinean naturalismand normative naturalism, in short, make for poor bedfellows.

2. Reductionisms

The words “reduction” and “reductionism” have been also been usedin a number of ways in philosophy of mind. One of the most familiar usesof the word in philosophy of mind makes “reductionism” a synonym for“type-type identity theory” – the view that each mental type stands in aone-to-one relation with a corresponding physical type (Place 1956). Typeidentity theory claims both an “upward” and a “downward” relationshipbetween mental and physical types. Given a mental type, one can infer“downwards” to physical type, and vice versa.

The type identity thesis has largely fallen out of favor in philosophy ofmind, due to concerns raised by Putnam (1980) about multiple realizationand the popularity of functionalist accounts of the mind. However, forpurposes of explanation, it is clearly only the “upward” relations thatmatter. The core of reductionism can be preserved in a way that allowsmultiple realization so long as fixing the physical type allows inferenceupward to a unique mental type as well. As a result, neither scientists,nor indeed all philosophers, have been inclined to give up on some sort of“reductionism” – i.e., one that is weaker than type identity – even if theyacknowledge the possibility of multiple realization for functional kinds.

Scientists, in particular, are often considerably more profligate in theirbestowal of the label “reduction” than are philosophers. Indeed, one canfind the term applied not only to very partial part-whole explanations(e.g., serotonin reuptake explains depression), and indeed to very differentforms of explanation such as what Kitcher (1981) calls “unifications”.Such a broad usage is probably not sufficiently precise as to be a propersubject of philosophical investigation (though see Bickle 1998), and iscertainly too weak to underwrite claims of metaphysical supervenience.

The core of the traditional philosophical notion of reduction wouldseem to consist in its being a form of explanation that has the followingdistinctive characteristics. First, it is explanation of a system in terms ofits parts – chemical reactions in terms of the properties of the atoms, forexample. Second, it is explanation in which, once one has an adequatetheory of the lower-level system in hand, one can then demonstrate thatall of the salient features of the higher-level system are a necessary conse-quence of these. This notion of reduction, which I have elsewhere termed“broad reduction” (Horst 2007), is a form of “conceptually adequate ex-planation” (Horst 1996).

This model of explanation was taken as canonical in two importantperiods in this history of philosophy of science. In the 17th century, it was

202 Horst

favored by innovators in early modern science like Galileo and Descartes,who viewed it on the model of geometric proof and construction, in whichone can begin with a conservative set of definitions and axioms and fromthese derive a rich set of theorems.

In the 20th century, it was favored by the Logical Postivists and Em-piricists, for whom the preferred model was not geometric construction butlogical syllogism. Both explanations within a science and inter-theoreticreductions were viewed by the Positivists as being deductive in form. Thismight not be evident from the form in which theories appear in sciencetexts, but could be obtained through the axiomatic reconstruction of par-ticular sciences and inter-theoretic reductions, a project that probablyreached its zenith in the work of Nagel (1961).

3. A Brief History of Recent Philosophy of Mind

Given this notion of reduction, we can locate a number of familiarphilosophical views with respect to the answers they give to four questions:

1. Is some form of inter-theoretic reduction the norm in the naturalsciences?

2. Do the objects, properties and events of natural sciences like chem-istry and biology metaphysically supervene upon those of physics?

3. Are mental phenomena like consciousness and intentionality reducibleto something non-mental?

4. Do mental phenomena metaphysically supervene upon physical phe-nomena?

When I was a graduate student in the 1980s, some combination ofreductionism and physicalism was still a kind of orthodoxy in philosophyof mind. Reductive physicalists characteristically answer yes to all fourquestions.1 But reductionists are not of one mind about the status ofreductionism itself. Some view it as a kind of second-order empiricalhypothesis: that it will turn out, in the end, that such inter-theoreticreductions will prove to be available. Others follow some of the LogicalPositivists in viewing reducibility as a kind of a priori norm guiding workin the sciences, and thus to provide a kind of litmus for the methodologicaland ontological legitimacy of the special sciences (particularly psychology)and their objects.

1Depending on just what one means by “reduction”, functionalists who are alsophysicalists might or might not qualify as reductionists as well. If “reduction” meanstype-type identity with physical or neural types, functionalists are not reductionists;but many functionalists do view their functional analyses as supplying a reduction ofmental kinds to functional kinds, each of whose instances is token-identical with aphysical particular.


As the twentieth century was entering its final quarter, a radical alter-native to reductionism came upon the scene in the form of eliminativism.Eliminativists generally agree with reductionists that reduction and su-pervenience are the norm in the physical sciences, but they hold that suchreductions are not available for mental phenomena, and that the properattitude to take towards such a gap is to conclude that the kinds of objectspostulated by “folk psychology” literally do not exist. They thus answer“no” to questions 3 and 4, though in the case of supervenience, the answeris “no” only in the trivial sense that nothing answering to the descriptionof mental states really exists at all, and hence cannot supervene uponanything.

Eliminativists also tend to agree with normative reductionists thatthere is an important relation between questions about inter-theoreticreducibility and questions of ontological legitimacy. A principled failureof reducibility in a special science like psychology would imply that itwas suspect, both methodologically and metaphysically, thus spawningthe need for a project of the “vindication of psychology.” As Fodor (1987,pp. 97, 98) put it,

The deepest motivation for intentional irrealism derives ... from acertain ontological intuition: that there is no place for intentionalcategories in a physicalistic view of the world; that the intentionalcan’t be naturalized.2 ...It’s hard to see ... how one can be a Realist about intentionalitywithout also being, to some extent or other, a Reductionist. If thesemantic and the intentional are real properties of things, it must bein virtue of their identity with (or maybe their supervenience on?)properties that are themselves neither intentional nor semantic. Ifaboutness is real, it must be really something else.

There were several voices in the 1970s and 1980s sounding the call forrecognition of an explanatory gap between mind and matter (Nagel 1974,Jackson 1982, Searle 1992, Levine 1983). However, this view did not re-ally re-enter the philosophical mainstream until the early to mid 1990s.3

Some important proponents of the gap, like Chalmers (1996) and Jackson(Chalmers and Jackson 2001), further claimed that a principled and abid-ing explanatory gap entails an ontological gap as well, thus supportingeither a substance or property dualism.

Most of these writers assume, along with reductionists and elimina-tivists, that inter-theoretic reductions and supervenience are the rule inthe physical sciences, and hold that that the mind is unique both in giving

2Fodor addresses this fear of intentional realism as a way of motivating his ownproject of naturalizing the mind, and thus dispelling the fear.

3I would identify watershed events here as the 1994 Tucson Conference on con-sciousness, and particularly the controversy around Chalmers’ presentation there andhis related book (Chalmers 1996).

204 Horst

rise to an explanatory gap and in failing to supervene upon things phys-ical. The explanatory gap, or in Chalmers’ terms “the hard problem ofconsciousness”, is thus seen to represent a unique and fascinating philo-sophical puzzle that cannot be addressed by ordinary scientific means.Some, but not all, advocates of the explanatory gap argue that an abidingand principled explanatory gap (as opposed to one that was merely reflec-tive of a current state of ignorance) entails a metaphysical gap as well, inthe form of a failure of the mental to supervene upon the physical. Theythus endorse what we might call a Negative Explanation-to-MetaphysicsConnection Principle:

Negative Epistemology-to-Metaphysics Connection (EMC):The irreducibility of A to B entails that A does not superveneupon B.

Some non-reductive physicalists likewise take the view that the mind isuniquely irreducible. Non-reductive physicalisms, however, had begun totake root in the 1980s for reasons unrelated to the explanatory gap, par-ticularly due to the influence of Davidson’s (1970) anomalous monism. Arelated view that emerged in the 1990s is the view sometimes called “mys-terianism”, associated with McGinn (1983, 1991). Mysterianism is at itsroot an epistemological view: that there is an explanatory gap in psychol-ogy, but that its roots lie in an inability of minds like ours to represent theexplanatory relations between body and mind. Mysterians generally arenon-reductive physicalists as well: they hold that mental properties su-pervene upon physical properties, but that this supervenience is abidinglyepistemically opaque to creatures with our particular representational ca-pacities.4 Mysterians and non-reductive physicalists thus answer “yes”to questions 1 and 2, along with everyone else so far; they answer “no”to question 3 (the psychological cannot be reductively explained by thephysical) but “yes” to question 4 (the psychological supervenes upon thephysical).

It is worth stressing that every position thus far surveyed has sharedcommon answers to the questions about the physical sciences: that thephenomena of sciences like biology and chemistry supervene upon those ofphysics, and that they can be explained by physical phenomena throughinter-theoretic reduction. The basic notion of mysterianism does, however,suggest a more radical possibility that one might wish to explore: namely,that the connections between the natural sciences might turn out to be

4The term “mysterian” is used in stronger and weaker senses. In its weaker sense,which I employ here, it is the view that some of the characteristically mental propertiescannot be explained in non-mental terms. In its stronger sense, it denotes the viewthat the mental is utterly ineffable, even in mentalistic terms. I should also notethat mysterianism does not itself commit those who hold it to supervenience or tomaterialism.


“gappy” and non-reductive as well. Such a possibility was hinted at inthe 1990s in some publications by Stich and Laurence (1994) and Baker(1995), but to the best of my knowledge was not extensively exploredwithin philosophy of mind before Horst (2007).

reductionists eliminativists dualists mysterians

reduction innatural science yes yes yes yes

supervenience innatural science yes yes yes yes

psychologicalreduction yes no no no

psychologicalsupervenience yes no∗ no yes (mostly)∗∗

positiveEMC yes yes yes yes

negativeEMC yes yes yes no

normativereduction yes/no∗∗∗ yes no no

Table 1: Summary of several positions discussed in the text withrespect to their interlevel relations. The assessment for mysteriansholds equally for non-reductive materialists. ∗ For eliminativists,the failure of supervenience is a trivial consequence of the claimthat there are no mental states to thus supervene. ∗∗ Davidsonand other interpretivists reject supervenience on the grounds thatthere are always multiple equally-good intentional characterizationsof a person’s behavior. ∗∗∗ Some reductionists, like the positivists,took reducibility to be a normative condition, while others, likeOppenheim and Putnam, took it only as a hypothesis.

Such a view is deserving of more exploration. For during the verytime period that discussions were going on in philosophy of mind, onthe assumption that inter-theoretic reduction is the rule in the physicalsciences, philosophers of science were moving in precisely the opposite di-rection. The reductionist view of the unity of science favored by many ofthe Positivists and Logical Empiricists received withering and even deci-sive criticisms, coming both from within Empiricism and from without.If there is a mainstream view about reduction in philosophy of sciencetoday, it is probably that reductionism, at least as conceived throughmost of the 20th century, was a failed project, motivated by a misguidedapriorism from Postivist philosophy of science, that the special sciences

206 Horst

are autonomous and self-justifying, and that there are in fact a variety oftypes of inter-theoretic relations to be found at the intersections of varioussciences, few of them amounting to inter-theoretic reduction.

As a result, although many conversations in philosophy of mind throughthe 1990s took themselves to be motivated by views in philosophy of sci-ence, the philosophy of science they were motivated by was more or lessthat of Oppenheim and Putnam (1958) or Nagel (1961), consisting ofviews that had already run their course within philosophy of science it-self. In short, recent philosophy of mind has been one of the last hidingplaces of philosophy of science of the 1950s.5

The question to be posed, then, is this: If we were to take into ac-count the rejection of inter-theoretic reduction as a norm for the specialsciences, what implications would this have for the mainline problematicin philosophy of mind? That is the question to be addressed in this paper.First, however, it is necessary to say a word about the waning fortunes ofreductionism in philosophy of science. While it is impossible to do morethan rough justice to the many strands of anti-reductionist philosophy ofscience, it is important to at least canvass their major points.

4. The Demise of Reductionism in Philosophyof Science

While everyone would agree that reductions are a particularly strongand useful form of explanation to have when one can get them, it is reallya second-order empirical question whether and where they are in fact to befound. Subsequent philosophers who attempted to extend Nagel’s projectof axiomatic reconstruction began to discover that mechanics turned outto be a special case, and almost unique in being susceptible to this formof reconstruction. Toulmin (1974, p. 610), for example writes

In mechanics – and in mechanics alone – the intellectual contentof an entire physical science could apparently be expounded as asingle mathematical calculus. Here was a complete natural sciencefree of logical gaps and incoherences ... . The temptation to holdtheoretical mechanics up as a mirror to other branches of science,and to demand that other sciences be construed on the same modeland achieve the same logical coherence, seemed irresistible. Yetthe very formal perfection of theoretical mechanics ought surelyto have ruled it out as the “type example” of a natural science,

5It is, sadly, not the only, or the best-publicized such hiding place. Mass-marketedbooks like Wilson’s (1998) Concilience and Crick’s (1993) Astonishing Hypothesis seemto make similar assumptions, and others of greater philosophical naivite. Wilson,for example, seems to think the greatest challenges to reductionism come from post-modernists and religious luddites; in fact they come from mainline analytic philosophersof science who have cut their teeth looking at the details of various actual sciences.


and prevented us from extrapolating conclusions about the “logicalstructure” of mechanics, so as to apply to natural sciences generally.Rather, we need to recognize how exceptional a science mechanicsreally is.

Patrick Suppes, who had himself undertaken the project of axioma-tizing a number of areas of science, likewise came to the conclusion that,while such reconstructions were possible in some areas of the sciences, theywere by no means the rule, and indeed mechanics was almost unique inthis regard (cf. Suppes 1974, p. 66). Meanwhile, historicist philosophersof science began to make an important turn away from the normative,aprioristic project of the “logic of science” and towards the hands-on in-vestigation of how science is practiced “in the wild”. Salmon (1971, 1984)and others showed decisively that some important methods of scientificexplanation were not syllogistic. Various writers through the 1980s and1990s noted that even when explanations could be reconstructed as syllo-gisms, these reconstructions came after the explanatory success had comethrough non-derivational forms (Schaffner 1967, 1974; Churchland 1986),and the reconstruction misdescribed theories “in the wild” (Nersessian1992, Craver 2002). And examination of what really takes place at theborders between theories revealed a much richer set of options, which wereexplanatorily useful and productive even when they fell short of reduction(Darden and Maull 1977, Bechtel and Richardson 1993).

Perhaps most damningly, case studies of purported inter-theoretic re-lationships have seemed to lead to the conclusion that true reductions areactually rather rare, and even the four or five much-touted examples donot deliver as advertised. As Silberstein (2002, p. 94) puts it,

Focus on actual scientific practice suggests that either there reallyare not many cases of successful epistemological (intertheoretic)reduction or that most philosophical accounts of reduction bearlittle relevance to the way reduction in science actually works. Mostworking scientists would probably opt for the latter claim. Oftendiscussed cases of failed or incomplete intertheoretic reduction inthe literature include:

1. the reduction of thermodynamics to statistical mechanics (Pri-mas 1991, 1998; Sklar 1999)

2. the reduction of thermodynamics/statistical mechanics to quan-tum mechanics (Hellman 1999)

3. the reduction of chemistry to quantum mechanics (Cartwright1997; Primas 1983)

4. the reduction of classical mechanics to quantum mechanics(such as the worry that quantum mechanics cannot recoverclassical chaos) (Belot and Eraman 1997)

208 Horst

Additional critique has arisen from philosophy of biology, denying thereducibility of evolutionary biology either to molecular genetics (Levins1968, Lewontin 1983) or to classical genetics (Kitcher 1984).

As noted earlier, a number of these writers would describe their ap-proach to philosophy of science as “naturalistic”. In philosophy of science,“naturalism” is a label that implies a rejection of any normative role givento standards imported from outside of the sciences themselves, such as thePositivist view of “the logic of science”. Such writers would insist thatit is only in conjunction with a study of actual scientific practice thatwe can determine not only (a) where (if anywhere) a particular type ofexplanation (e.g., a reduction) is to be found, but also (b) what sorts ofstandards of explanation are appropriate to the several sciences.

This is, of course, only a cursory overview of several broad trendswithin philosophy of science. It is impossible in the course of an articlelike this to provide much more detail on the arguments for each position,much less to adjudicate questions on which there are remaining controver-sies.6 In the context of this article, this exposition is largely preparatoryto consideration of questions in philosophy of mind. To put it briefly, ifone accepts these anti-reductionist results in philosophy of science, whatimplications would this have for how one does philosophy of mind? Andin particular, to what extent would some familiar problematics be trans-formed or undone?

5. Reductive and Eliminative Physicalisms

In philosophy of mind, at least one moral seems quite straightforward:Post-reductionist philosophy of science is bad news for reductive physi-calism. It was bad enough for the reductive physicalist that there seemto be explanatory gaps with respect to intentionality and consciousness.But as long as those gaps were confined to psychology, the appearance ofsuch gaps could be offset by the sense that that appearance might be –indeed, must be – mistaken, given what we know about the natural world.

If nature is generally united by reductive relations, it seems reasonableto conjecture that reductive explanations will eventually be found for allmental phenomena as well. But if there are “explanatory gaps all the waydown”, no such assurance is justified. Indeed, if explanatory gaps in theforms of failures of reduction are the rule rather than the exception in eventhe natural sciences, expecting psycho-physical reductions would meanexpecting closer relations between brain and mind than are found betweenphysics and chemistry. Such a hope might pan out, but its adherents arebetting against long odds.

6I would commend the Blackwell Guide to Philosophy of Science (Machamer andSilberstein 2002) chapter on “Reduction, emergence and explanation” as a startingplace for those interested in reading further on these topics.


Post-reductionist philosophy of science is equally hard on elimina-tivism. If inter-theoretic reductions are rare in the sciences generally, it isnot merely beliefs and desires that would be in need of elimination, but allof the irreducible entities of biology, chemistry and even non-fundamentalphysics. There are indeed a few metaphysicians who are willing to endorsethe idea that the only things that really exist are the simplest entities offundamental physics, but this is far more radical a conclusion than evenmost eliminativists are prepared to accept. On the other hand, if we arewilling to accept the reality of irreducible entities in other special sciences,it is hard to see why we should not do so in the case of psychology as well.

6. The Explanatory Gaps and Metaphysics

At first sight, then, post-reductionist philosophy of science looks likegood news for proponents of the explanatory gap. It undermines much ofthe unease that might be felt about a unique gap in the case of the men-tal, which has been for many an important defeater for their intuitionsthat there is such a gap. But on the other hand, if it is “gaps all the waydown”, the psychological gap ceases to be such a uniquely interesting andsexy problem as well. And in particular, it puts the dualist in an awkwardposition. For one important argument for dualism is based on the gap,plus what I earlier called the “negative explanation-to-metaphysics con-nection principle” (or Negative EMC Principle). This is the idea that aprincipled explanatory gap entails a metaphysical gap as well in the formof a failure of supervenience.

Dualists have argued that there is no such gap within the natural sci-ences, hence only one physical substance, but a unique gap in psychology,entailing exactly one additional substance, for a grand total of two (orthree if, like Descartes, you number God as a third). But if it is gapsall the way down, Negative EMC does not entail dualism, but a radicalontological pluralism. And radical such a pluralism would seem to be: ifabiding and principled gaps imply failures of supervenience, and chem-istry and biology are not reducible to physics, then they do not superveneupon physics either (and likewise, mutatis mutandis, for statistical me-chanics/thermodynamics, etc.). This is a result that few will find easy toswallow.

There is, however, also an intuition that many experience here that,while there are explanatory gaps outside of psychology, the psychologicalgaps are in some way special, and they are special in a way that carriesdifferent metaphysical consequences than the physics-chemistry gap. Ifthis is so, it may be possible to avoid radical ontological pluralism even ifwe accept theory pluralism in philosophy of science. I do think there arespecial issues about explaining experience in non-experiential terms that

210 Horst

do not have parallels in, say, (partially) explaining atomic-level propertiesin terms of particle-level properties or living systems in terms of theirproper parts, and these special problems are central to the motivatingconcerns (though not always the arguments) of advocates of the mind-brain gap like Levine and Chalmers.

I shall provide one account of why these issues arise, and why they arespecial, in Section 9. I have been using the expression “explanatory gap”as a more compact way of saying “(principled) failure of reductive expla-nation”, which is not meant to imply that all failures of reductive expla-nation fail for the same reasons.7 But the arguments about reductionism,dualism, and eliminativism that I have been considering are argumentsabout the consequences of reducibility and irreducibility per se, and notabout underlying sources of irreducibility. Mere irreducibility, whateverits source, has clear consequences for reductionism and eliminativism, andI shall claim that non-reductive materialism needs to find another moti-vation for its materialism is justified to the extent that the plausibility ofmaterialism was largely motivated by an assumption of reducibility.

By contrast, the case for widespread irreducibility leaves an open doorfor the dualist: to argue that some further feature that makes mentalphenomena irreducible can provide the basis for a better argument fordualism because it leaves us with an explanatory gap that is different inkind from those that are “gaps” merely in the sense of failures of reducibil-ity. But this would need to be an argument that does not depend upon thepremise that irreducibility itself is a sufficient condition for metaphysicalindependence. Any case for dualism that is not a better case for pluralismmust not rest on irreducibility, but on something else.

In the following sections I shall examine three metaphysical possibili-ties: materialist, dualist and pluralist. I shall also mention in passing twoadditional strategies for dealing with the problems raised which I shallnot dwell upon here. One of these is to simply hold out for a day whenreductionism will make a stunning comeback. The other is to disavow themetaphysical side of philosophy of mind altogether, and to concentrateonly on “hands-on” philosophy of the cognitive sciences. I propose to ig-nore these options in the rest of the paper, but do so without prejudiceto their possible virtues.

6.1 Non-Reductive Physicalism and Mysterianism

There is nothing about mysterianism or non-reductionism that pre-vents its advocates from holding that it is true all the way down. Theorypluralism may provide a challenge to materialism and a monistic meta-physics, but it is not clear that it presents a refutation; and indeed the

7Indeed, I think there are probably a number of different types of reasons in differentcases. To adapt a quote from Tolstoy, each unhappy family is unhappy in its own way.


non-reductive physicalist might actually turn this to his advantage. If, say,chemistry is not reducible to physics, yet we are sure that it supervenesupon physics, this gives us good reason to assume that supervenience iscompatible with the inavailability of reductions. And if it is compati-ble in one case, it should be compatible in all cases, including that ofpsycho-physical supervenience. One might even be inclined to say thatnon-reductive physicalism is in a better evidential position if one embracestheory pluralism.

There is something a bit illusory about this inclination, though. Thiscan be seen by noting that the strength of the non-reductive physicalist po-sition depends on the direction from which one arrives at it. Historically,the popularity of materialism came hand-in-hand with the popularity ofreductionism. Indeed, what were taken to be reductive successes in thesciences were precisely what led people like Hobbes and Laplace to be ma-terialists. Where there are reductions, supervenience comes at no extracharge. And if reductions are shown to be unavailable, it is natural for thephysicalist to look to non-reductive physicalism as the most conservativefallback position.

But suppose we start the other way: we start with irreducibility ormysterianism, and then ask: What is the best metaphysical interpretationof irreducible correlations? Now there seems to be little reason, otherthan some sort of philosophical taste (e.g., for the simplest ontologicalinventory) to prefer the view that an A-B correlation is grounded in ametaphysical necessity (that B → A in all possible worlds), over the viewthat it is merely a lawlike and/or causal connection.

The non-reductive side of non-reductive matieralism may be based onstrong evidence; but the materialist side is more like a standpoint of faith.There is nothing wrong with standpoints of faith, but this deprives mate-rialists of the rhetorical high ground aligning them with both reason andevidence and their opponents against it. Without reduction, materialismloses such grounding as it might have had in scientific explanation.

6.2 Dualism

The case is similar with respect to dualism. The dualist can acceptthe view that there are explanatory gaps all the way down. What hecannot accept is that they all entail metaphysical gaps, upon pain ofceasing to be a dualist by counting to three and beyond. There is, ofcourse, a clear agenda here for the dualist: explain why the psychologicalgaps (for intentionality and consciousness) are different from the gapsin the natural sciences, and why this difference entails a difference inmetaphysical import as well. For this writer, the intuition that there issuch a difference is very strong, though of course that does not decidewhether this indicates truth or the grip of a powerful illusion.

212 Horst

But there is a great cost to the dualist in making such a move. For to doso, she must give up the Negative EMC Principle. And this principle hasbeen the linchpin of one of the key (perhaps the key) argument employedby dualists to establish their own position. The main reason that writerslike Chalmers and Jackson (of a previous decade, at least) favor dualism ison the basis of the explanatory gap, combined with the view that abidingand principled explanatory gaps entail ontological gaps as well.

But if it is gaps all the way down, it would seem there is a forced choicehere between abandoning dualism in favor of a more radical pluralism, orelse abandoning the Negative EMC. To do the former is to cease to be adualist; to do the latter is to retain dualism at the cost of giving up theprime evidence for it, thus making it also more of a standpoint of faithor of philosophical taste. And this applies, mutatis mutandis, to propertydualism: if explanatory gaps entail distinct categories of properties, andthere are many explanatory gaps, there are many distinct categories ofproperties.

6.3 Pluralism

We have thus come to the surprising conclusion that, without reduc-tions in the natural sciences, both dualism and materialism are harder tomotivate. It is worth exploring the pluralist alternative. This involves atleast a pluralism about theories or explanations – a kind of disunity ofscience view, at least insofar as the kind of unity being denied is inter-theoretic reduction. What does it involve on the metaphysical side? Therewould seem to be at least two types of alternatives.

One we might call “realist pluralism”: roughly, the view that there isan irreducible plurality of ontological kinds, numbering at least as manyas the theoretical posits of irreducible (true) scientific theories. This isperhaps the view we find in Dupre (1993), who argues that disunity ofscience is a result of a prior ontological disunity (as reflected in the titleof his book The Disorder of Things.)

But one might alternatively take a cognitivist or pragmatist or idealistapproach to disunity, holding that failures of supervenience are ultimatelyan artifact of something about our cognitive architecture, or the inter-ests with which we frame our several scientific theories, or the apparatusthrough which we interact with the world. Such views differ from bothdualism and materialism, not at the level of “inventory ontology” (the listof things they hold to exist) but at the level of “critical ontology” – theaccount they give of what it is to be a thing in the first place.

Pragmatists, idealists and cognitivists are inclined to characterize bothdualism and materialism as versions of “naive realism” – the view thatthe world unproblematically divides itself up into a canonical list of ob-jects and properties in a mind-independent way – and propose their own


respective accounts of objecthood. This sort of metaphysics might be ap-plied to our problems as well, and indeed I will recommend one version ofit, which I call “Cognitive Pluralism”.

6.4 Realist Pluralism

The realist pluralist takes irreducibility of an object or property tobe a mark of its being fundamental. On such a view, a radical theorypluralism on the epistemic side entails a radically pluralistic inventory onthe ontological side. (It might be possible to distinguish the thesis thatirreducibility implies fundamentalness from the Negative EMC, but onlyat the cost of divorcing ontology from scientific realism.)

To the extent that one is wedded to this sort of realism and “fundamen-talism”, this might be an attractive position. However, its metaphysicalcosts are both great and apparent. On the one hand, it is extremely onto-logically profligate. Instead of one basic ontological kind, or even a hand-ful, it ends up committed to hundreds or even more. On the other hand,at least some of its denials of supervenience seem quite suspect. Short ofevidence that some particular chemical property is underdetermined byphysical properties, we are generally wont to assume that fixing the phys-ical properties for a possible world thereby fixes the chemical propertiesas well. This intuition might be defeasible in the face of possible empiricalor philosophical argumentation, bit it dies hard. There is, in short, anunstable triad of views here:

(1) the Negative EMC Principle,

(2) theory pluralism in philosophy of science,

(3) the intuition that the phenomena of natural sciences are metaphysi-cally determined by the complete set of physical facts and laws castat the level of the simplest physical objects.

The realist pluralist rejects (3), but it is by no means clear that this isthe least costly move to make.

7. Cognitive Pluralism

As an example of a non-realist pluralist approach, I shall briefly de-velop a view that I call “Cognitive Pluralism” (for a longer treatment,cf. Horst 2007, 2011). The Cognitive Pluralist approaches scientific dis-unity in cognitive and pragmatic terms. To have a scientific theory is not,after all, simply to passively reflect the world like Rorty’s (1981) mirrorof nature, but to actively model specific features of it, one (or a few) at atime.

Modeling features of the world, in turn, involves both abstractionsand idealizations. Abstraction involves screening out some features of the

214 Horst

world to bring others into focus, like viewing an object through a partic-ular type of lens. Idealization (as I shall stipulatively use that term) is aspecial type of abstraction whose characteristic feature is that it screensout features that matter in vivo, thus creating a gap between behavior-as-reflected-by-the-model and real world kinematics. Thus, for example, agravitational theory might be idealized in several ways: qua gravitationaltheory, it models only the gravitational aspects of kinematic problems, andbrackets contributions of, say, electromagnetism. And thus, as Cartwright(1983) has pointed out, objects never behave exactly as gravitational the-ory, taken alone, would predict.8

In addition to this kind of “bracketing idealization”, the model mightalso make “distorting idealizations”, such as treating objects as point-masses. This is comparatively innocent in some cases, but ceases to beinnocent when factors like aerodynamics are also in play. (A sheet of papercrumpled into a ball falls quite differently from a similar sheet folded intoa paper airplane.)

A model also involves some particular representational system – aguiding metaphor, for example, and a choice of mathematical tools (aparticular geometry or topology). Thus wave and particle models capturedifferent aspects of the phenomena they model by employing differentrepresentational systems. What representational systems human mindscan employ is, further, constrained by human cognitive architecture, andthe models actually employed then constrain the kinds of information thatcan be captured and exploited as opposed to what is screened out. Toidealize and apply a particular mathematical structure to a messy real-world situation is necessarily to lose information as the price paid forinsight and computational tractability.

Given that scientific models are idealized, optimized for particularpragmatic goals, and subject to constraints of cognitive architecture, thiscan explain at least some types of failures of reducibility. Not all repre-sentational systems are in fact reducible to a common denominator. Theycan, nonetheless, be models of the same reality, as in the case of parti-cle and wave models in classical optics or quantum mechanics, much asone can take pictures of a single object from different angles and throughdifferent camera lenses. However, pasting together such snapshots doesnot produce, and could never produce, a master picture of how the objectlooks from all angles at once. Pictures do not work like that, and neither,quite plausibly, do theories or models.

The Cognitive Pluralist might further be inclined to view this featureof scientific disunity as a special case of a broader phenomenon that is in-dicative of human cognitive architecture. Mind and brain display a certain

8And hence we must view laws as stating something other than universal claimsabout how objects actually behave if we are to preserve the intuition that they aretrue; cf. Horst (2011).


amount of division of labor into special-purpose, task-optimized systems,even at the hardware level of the brain. Special-purpose mechanisms thatare species-typical and have characteristic neural localizations are oftenreferred to as “modules” (Fodor 1983). Domain-specific reasoning, whileprobably not accomplished by a partitioning of brain hardware, seems tobe an analogous phenomenon involving learned models rather than hard-wired modules.

In both cases, models are driven by pragmatic constraints (via learningor natural selection respectively) and subject to prior constraints imposedby cognitive architecture. Scientific models are idealized, local, and prag-matically constrained because they are a product of human minds andbrains, and that is the kind of models human minds and brains are de-signed for. And, as a consequence, the question of whether a theory A canbe reduced to B turns out to be an empirical question, not just about howthe world is, but also about the capabilities of the mind itself, and how itrepresents A and B. This has led us from the cognitivist side of the storyto the pluralist side: the hypothesis that scientific disunity, like domainspecificity of reasoning and brain modularity, is not just a symptom ofthe immaturity of our scientific knowledge, but a necessary by-product ofhow our minds are designed (cf. McGinn 1991, Pinker 1997).

As a purely epistemic claim, Cognitive Pluralism might be acceptableand even welcome to dualists and especially to materialist mysterians. Itprovides a model on which failures of reduction need not entail failures ofsupervenience, if only by providing an alternative explanation of scientificdisunity. The monist only wants reality to be unified; if it is only reality-as-represented-by-us that is disunified, a unified world need not be indanger.

It seems to me, however, that this move is bought at considerable cost.For if the materialist adopts a cognitivist or pragmatist slant on scientifictheories, and thereby on the entities postulated by those theories, she isfaced with the choice between skepticism and idealism that plagued theBritish Empiricists. Either the “matter” one is talking about is the sortof thing that appears in scientific theories, or it is something else on theorder of a Lockean real essence or Kantian Ding an sich. If it is the former,then “the world” really is disunified, because it consists in the patchworkof objects and properties found in the sciences, whose critical ontology liesat the interface between our minds (augmented by regimented practicesof experimentation and mathematical tools) and ultimate reality. “Theworld” here is, in Kantian terms, the phenomenal world, and is a possibleobject of knowledge, but apparently not unified knowledge. But if “theworld” is more on the order of noumena – something other than reality-as-represented-in-our-theories – then it is unknowable to us, or at leaststarkly separated from our theories, thus leading to a Lockean form ofskepticism.

216 Horst

The metaphysical side of Cognitive Pluralism is to embrace the kindof view that Kant recommended: that we preserve the possibility of scien-tific knowledge by treating it as knowledge of phenomena – albeit perhapsphenomena that we play a role in creating through regimented scientificexperimental practice (cf. Hacking 1983). Here the lines between “real-ism” and “antirealism” are to my mind very thin – perhaps so thin as tobe meaningless. (The real issue may be closer to what Cartwright (1994)calls “fundamentalism” rather than realism.) For the Cognitive Pluralistneed not (and indeed should not) take the view that all ways of modelingthe world are on a par with one another. Some models are more apt thanothers; and while cognitive architecture and pragmatic interests shapethe space of possible models, it is experiment that settles the questionof which of these are more and less apt. To fail to do this is to mistakedaydreaming for science.

Moreover, scientific models can even be epistemically privileged overother types of models, in that they are products of a kind of regimentedempirical practice that excludes as much as possible of what are arbitrarybyproducts of human interests and cognition, and insists upon settingthings up so that the world itself decides as many crucial questions aspossible.

8. Cognitive Pluralism, Supervenience,and Modal Intuitions

What, then, should the Cognitive Pluralist say about the two issuesthat caused problems for other views: the implications for ontology andthe apparent specialness of the psychological gaps? The ontological prob-lem was framed in terms of supervenience, and this is a notion thatthe Cognitive Pluralist might regard with some suspicion. In particular,he ought to be concerned about standard formulations of superveniencewithin a modal metaphysics cashed out in terms of a possible worlds se-mantics (PWS). Consider how one gets at the notion of a possible world.One starts with a set of all propositions or a set of all states of affairs.A world is then conceived of as being, or as being defined by, a mappingfrom this set onto truth values. A possible world is one in which there areno inconsistencies generated by such a mapping.

Now ought a Cognitive Pluralist to be happy with such a project?Well, perhaps at the level of it being one more conceptual tool that can dosome limited amount of useful work.9 But if it is supposed to give insightinto the deep structure of metaphysical reality, things look a bit suspicious.

9And this is not an unimportant point, as it avoids the more principled anti-metaphysical views expressed by Ladyman et al. (2007) and Chemero and Silberstein(2008).


On the one hand, it is suspicious from a cognitivist standpoint. To startoff speaking of “the set of all properties” (“all propositions”, “all states ofaffairs”) would seem to be exactly the kind of naive realist move that theCognitive Pluralist, qua cognitivist (or similarly the idealist, pragmatist,or social constructivist) takes issue with, as it assumes that there is someunique, canonical, mind-independent way that reality divides itself up intoproperties, propositions or states of affairs. The cognitivist may agree thatthere are enterprises in which one has to think like a naive realist.10 Butthe juncture at which one decidedly must avoid this impulse is when oneis trying to do fundamental metaphysics.

The Cognitive Pluralist also has a problem with PWS qua plural-ist. For PWS seems to assume that the various things we can representthrough local models can be united into a single master system. PWSneed not imply reductive unification – propositions can be consistent withone another without being reducible to a common denominator. But evenhighly apt idealized models can sometimes generate contradictions, as isin fact the case with general relativity and quantum mechanics. The Cog-nitive Pluralist rejects the view that such a circumstance implies that atleast one of the models is false or inapt.

Indeed, each may deserve honorifics like “true” or “apt” as much asanything deserves them. And yet they can generate contradictions, dueto the ways they are idealized and partial. If you want a kind of “truth”that entails consistency too, you need to work with unidealized theories.Perhaps God can get at this sort of truth; but, the Cognitive Pluralisthypothesizes, the human mind cannot. But if two theories T1 and T2

are true of a world W, yet generate contrary predictions in some casesin W, then either W is not a possible world as that notion is generallyunderstood (as that requires consistency) or else true theories do notimply their predictions. (Indeed, one could even derive the conclusionthat the actual world is not a “possible world” in the sense required byPWS.)

What moral should one draw from this? I think the deep issue here isthat scientific laws cannot be smoothly grafted onto the kind of semanticmodel represented by PWS, and may not be easily integrated with modalmetaphysics at all. From the Cognitive Pluralist standpoint, this does notimply that either enterprise (scientific modeling or modal metaphysics) isindividually problematic, as Cognitive Pluralism holds that in generaldisparate models may not be integratable with one another. But it doesseem to present problems for enterprises that require us to mix them

10Notably, in order to assess the fidelity or aptness of a representational system, onemust compare it to an independent description of the world that is treated as canonical.The epistemologist may know full well that this description, too, is idealized, but forpurposes of the task of fidelity-assessment treats it as though it described how theworld is in itself.

218 Horst

together. And speculating on whether the truths of one science superveneupon those of another would seem to be just such an enterprise.

More generally, a Cognitive Pluralist should regard the use of intu-itions in metaphysics with a certain degree of caution. What is the sourceof intuitions? One natural proposal, given the model-based cognitivismdescribed here, would be that things seem intuitively necessary when theyare required by the rules of a particular mental model, and intuitively im-possible when they violate the rules of a particular model. (We might lookfor the sources of abiding paradoxes in pairs of models with conflictingrules.) But of course the intuitions that arise from a model are only asapt as the model itself, and models are generally idealized, and not aptlyapplied in all circumstances. But metaphysical intuitions generally speakto matters where we are concerned with absolute generality.

Likewise, we might expect intuitions arising from models that havewithstood the test of time in particular uses within their own domainsto be more trustworthy than not. But it seems unlikely that modelsthat have been products of natural selection or a long learning historyare aimed at metaphysical questions, nor that they would be reliable intracking modal facts. Selection and learning are sensitive to empiricalregularities, but not to modal facts that outstrip empirical regularities.But modal metaphysical questions turn precisely on things that go beyondempirical regularity, and hence to be things for which there is unlikely tobe a mechanism for tuning a model to aptness, and hence to providetrustworthy intuitions.

9. Cognitive Pluralism and the Special Statusof the Mind-Brain Gap

What about the intuition that the explanatory gaps in psychologyare wider and deeper than those in the natural sciences? The cognitiviststrain of the brand of pluralism I am recommending does have special re-sources for addressing this. Transcendental idealists like Kant and Husserlhave already pointed out that there are special problems in trying to talkabout subjective experience as though it were itself a thing in the world,related to other things in the world. On the one hand, a description ofsubjectivity seems to require a peculiar representational form, in whichself/thought/object are related in a single intentional state, not as threeobjects in an objective relation, but rather as three moments of a singleexperience. Husserl additionally points out that to take either a thoughtor the transcendental subject and treat it as a thing (an object) is neces-sarily to distort it as it appears in lived experience. If this is correct, thenthere are problems in trying to create a mind-body theory (or better, anexperience-body theory) that do not arise when relating two objects orsystems of objects.


Further, for idealists and pragmatists, “ontology”, in the sense of theinventory of the world, is not metaphysical bedrock. There is a deeperenterprise of examining what it is to exist, or to be a unity. And this theycash out in terms of relation to a subject that thinks or has theoreticaland practical interests. The point here is not that the inventory of theuniverse is first and foremost one of conscious subjects and that physicalphenomena are causally or compositionally derivative from these. Thepoint is not about inventory at all. But there is a deeply anti-reductionistmoral here stemming wholly from the cognitivist/idealist/pragmatist side.And that is that objecthood itself needs to be cashed out in terms ofcognition. What it is to be an object is to be the sort of thing that canbe made a possible object of cognition.

Objecthood is understood in terms that relate it to thinking subjects.And thus this account of objecthood is one in which subjecthood is insome sense “already on the scene” and “prior” to the objects of cognition.The priority is more of the logical than the causal or temporal sort, butit is an important sort of priority nonetheless, as it implies that at somelevel subjectivity cannot be exhaustively explained in objective terms, bethey natural-scientific or otherwise.

How plausible one finds such a position will, I suspect, depend verylittle on one’s attitude towards the pluralist strand and very much on one’sindependent attitude to the costs and benefits of cognitivism, pragmatismor idealism as compared with naive realism. (And of course, those whofavor the latter generally do not call their view “naive realism” in thefirst place.) However, if the earlier parts of this article are correct, thetraditional arguments in favor of “realist” ontologies like dualism andphysicalism in fact turned quite a bit upon assumptions about reducibilityand its relation to metaphysical supervenience.

Theory pluralism takes away key arguments for both dualism andphysicalism, and indeed if traditional assumptions about the negativeEMC are retained, it leads to a radical ontological pluralism. Whetherone makes the non-realist turn or not, then, it seems necessary to posita greater gulf between science and metaphysics than most dualists andmaterialists have (with notable exceptions such as Locke) heretofore as-sumed. And with this in mind, the barriers to the cognitivist turn seemsignificantly reduced. Add to this that cognitivist story is one on whichanswers to several puzzles simply fall out – why there should be scientificdisunities, why the psychological gap is special, and how to avoid radicalontological pluralism – and it would seem to be a strong contender andworthy of our attention.

220 Horst

10. Is Cognitive Pluralism Naturalisticor Anti-Naturalistic?

What is the relationship between Cognitive Pluralism and naturalism?If we mean “naturalism” in the roughly Quinean sense in which the wordis used in epistemology and philosophy of science, Cognitive Pluralism iscertainly compatible with naturalism. Indeed, to the extent that Cog-nitive Pluralism is concerned with empirical hypotheses about how themind works and their philosophical implications, or is engaged with theanalysis of things like the disunities found in the sciences, what I havepresented is a “naturalistic” research program.

If, however, we are concerned with the metaphysical sense of “natural-ism”, we must separate Cognitive Pluralist epistemology from CognitivePluralist metaphysics. The epistemology is one that, as I have indicated,some naturalists (particularly non-reductive physicalists) might well findinitially attractive. As a metaphysical position, however, Cognitive Plu-ralism would seem to be anti-naturalistic in much the same ways as aretranscendental idealism or pragmatism. These are all positions that rejectthe assumption that, when we have found the most fundamental posits ofnatural science, we have reached philosophical bedrock as well: there is afurther, and indeed a deeper, story to be told about what it is to be anobject, even an object of a fundamental physical kind.

It is quite possible that there are some philosophers who consider them-selves to be both naturalists and either cognitivists or pragmatists, andthey may wish to contest my restriction of the word “naturalism” to viewsthat are committed to naive realism and reject critical metaphysics. Thiswould be a conversation worth having, though I do not believe that I haveyet had it.

11. Conclusion

Philosophy of mind would seem to be at a crossroads. Familiar prob-lematics have tended to be predicated upon the assumptions (1) that thereare widespread reductions in the natural sciences, (2) that there are closerelations between explanation and metaphysics, and (3) that it is safe totreat these problems from the standpoint of a naively realistic inventoryontology without recourse to critical ontology.

The first assumption seems to have turned out to be empirically false asa second-order claim about the relationships between theories and modelsin different scientific domains. This, in turn, challenges the evidential sta-tus of all of the familiar positions. Reductive and eliminative physicalismare damaged, perhaps past resuscitation. Without reductions, material-ism is left without the primary means of arguing that inter-domain cor-


relations are symptoms of metaphysical supervenience rather than someweaker relation, and hence even non-reductive materialism is set back.

But the dualist alternative also loses argumentative ground as it musteither yield arguments based on the Negative EMC (in which case it too isill-motivated) or else admit that there are failures of supervenience outsideof psychology (in which case it ceases to be dualism and becomes a moreradical pluralism). Moreover, the explanatory gap in psychology seemsno longer to be a unique phenomenon; and if it is not unique, one mightraise the question of whether it is as interesting or important as we mighthave supposed as well. There is, to be sure, an intuition to the effectthat there is something different about the psychological gaps, but thisintuition needs to be fleshed out in greater detail before it can be put tophilosophical use.

The main suggestion of this article is therefore that post-reductionistphilosophy of science forces us to rethink some mainline problems andpositions in the philosophy of mind. To this I have added an outlineof what I take to be a promising way of proceeding: by re-examiningphilosophy of mind from the Cognitive Pluralist standpoint, which holdstheory pluralism to be a product of how the mind conceives of the world,and to explore the cognitivist turn in metaphysics as an alternative tonaively realistic inventory ontology.

References

Baker L.R. (1995): Explaining Attitudes, Cambridge University Press, Cam-bridge.

Bechtel W. and Richardson R.C. (1993): Discovering Complexity: Decomposi-tion and Localization as Strategies in Scientific Research, Princeton UniversityPress, Princeton.

Belot G. and Earman J. (1997): Chaos out of order: Quantum mechanics,the correspondence principle and chaos. Studies in History and Philosophy ofModern Physics 28(2), 147–182.

Bickle J. (1998): Psychoneural Reduction, the New Wave, MIT Press, Cam-bridge.

Cartwright N. (1983): How the Laws of Physics Lie, Oxford University Press,Oxford.

Cartwright N. (1994): Fundamentalism vs. the patchwork of laws. Proceedingsof the Aristotelian Society 94, 279–292.

Cartwright N. (1997): Why physics? In The Large, the Small and the HumanMind, ed. by R. Penrose, A. Shimony, N. Cartwright and S. Hawking, CambridgeUniversity Press, Cambridge, pp. 161-168.

Chalmers D. (1996): The Conscious Mind: In Search of a Fundamental Theory,Oxford University Press, Oxford.

222 Horst

Chalmers D.J. and Jackson F. (2001): Conceptual analysis and reductive ex-planation. Philosophical Review 110(3), 315–361.

Chemero A. and Silberstein M. (2008): After philosophy of mind: Replacingscholasticism with science. Philosophy of Science 75, 1–27.

Churchland P. (1986): Neurophilosophy. Toward a Unified Science of the Mind-Brain, MIT Press, Cambridge.

Craver C.F. (2002): Structure of scientific theories. In Blackwell Guide to thePhilosophy of Science, ed. by P. Machamer and M. Silberstein, Blackwell, Ox-ford, pp. 55–79.

Crick F. (1993): The Astonishing Hypohtesis: The Scientific Search for the Soul,Scribner, New York.

Darden L. and Maull N. (1977): Interfield theories. Philosophy of Science 44,43–64.

Davidson D. (1970): Mental events. In Experience and Theory, ed. by L. Fosterand J. Swanson, University of Massachusetts Press, Amherst, pp. 79–101.

Dupre J. (1993): The Disorder of Things: Metaphysical Foundations of theDisunity of Science, Harvard University Press, Cambridge.

Fodor J.A. (1983): Modularity of Mind: An Essay on Faculty Psychology, MITPress, Cambridge.

Fodor J. (1987): Psychosemantics: The Problem of Meaning in the Philosophyof Mind, Bradford, Cambridge.

Hacking I. (1983): Representing and Intervening: Introductory Topics in thePhilosophy of Natural Science, Cambridge University Press, Cambridge.

Hellman G. (1999): Reduction(?) to what? Comments on L. Sklar’s “Thereduction(?) of thermodynamics to statistical mechanics. Philosophical Studies95 (1-2), 203–214.

Horst S. (1999): Evolutionary explanation and the hard problem of conscious-ness. Journal of Consciousness Studies 6(1), 39–48.

Horst S. (2007): Beyond Reduction: Philosophy of Mind and Post-ReductionistPhilosophy of Science, Oxford University Press, New York.

Horst S. (2009): Naturalisms in philosophy of mind. Philosophy Compass 4(1),219–254.

Horst S. (2011): Laws, Mind and Free Will, MIT Press, Cambridge.

Husserl E. (1913/1931): Ideen zu einer reinen Phanomenologie und phanomeno-logischen Philosophie, Nijhoff, The Hague. English edition: Ideas: General In-troduction to Pure Phenomenolgy, transl. by W.R. Boyce Gibson, Collier Books.

Jackson F. (1982): Epiphenomenal qualia. Philosophical Quarterly 32, 127–136.

Kitcher P. (1981): Explanatory unification. Philosophy of Science 48(4), 507-531.

Kitcher P. (1984): 1953 and all that: A tale of two sciences. PhilosophicalReview 93, 335–373.

Ladyman J., Ross D., Spurrett D., and Collier J. (2007): Everything Must Go:Metaphysics Naturalized, Oxford University Press, Oxford.


Levine J. (1983): Materialism and qualia: The explanatory gap. Pacific Philo-sophical Quarterly 64, 354–361.

Levins R. (1968): Evolution in Changing Environments, Princeton UniversityPress, Princeton.

Lewontin R. (1983): Biological Determinism. Tanner Lectures on Human Val-ues Vol. 4, University of Utah Press, Salt Lake City, pp. 149–183.

Machamer P. and Silberstein M., eds. (2002): Blackwell Guide to the Philosophyof Science, Blackwell, Oxford.

McGinn C. (1983): The Subjective View, Oxford University Press, Oxford.

McGinn C. (1991): The Problem of Consciousness: Essays Toward a Resolution,Blackwell, Oxford.

Nagel E. (1961): The Structure of Science, Harcourt Brace, New York.

Nagel T. (1974): What is it like to be a bat? Philosophical Review 83(4),435-450.

Nersessian N.J. (1992): How do scientists think? Capturing the dynamics ofconceptual change. In Cognitive Models of Science. Minnesota Studies in thePhilosophy of Science II, ed. by R.N. Giere, University of Minnesota Press,Minneapolis, pp. 3–45.

Oppenheim P. and Putnam H. (1958): The unity of science as a working hy-pothesis. In Concepts, Theories and the Mind Body Problem. Minnesota Studiesin the Philosophy of Science II, ed. by H. Feigl, M. Scriven and G. Maxwell,University of Minnesota Press, Minneapolis, pp. 3-36.

Pinker S. (1997): How the Mind Works, Penguin, London.

Place U.T. (1956): Is consciousness a brain process? British Journal of Psy-chology 47(1), 44–50.

Primas H. (1991): Reductionism: Palaver without precedent. In The Problemof Reductionism in Science, ed. by E. Agazzi, Kluwer, Dordrecht, pp. 161–172.

Primas H. (1983): Chemistry, Quantum Mechanics, and Reductionism, Springer,Berlin.

Primas H. (1998): Emergence in the exact sciences. Acta Polytechnica Scandi-navica 91, 83–98.

Putnam H. (1980): Brains and behavior. In Readings in the Philosophy ofPsychology, ed. by N. Block, Harvard University Press, Cambridge. Originallypublished in Program of American Association for the Advancement of Science,Section L (History and Philosophy of Science), December 27, 1961.

Quine W.V.O. (1969): Epistemology naturalized. In Ontological Relativity andOther Essays, Columbia University Press, New York, pp. 69–90.

Quine W.V.O. (1992): Pursuit of Truth, Harvard University Press, Cambridge.

Rosenberg A. (1996) : A field guide to recent species of naturalism. BritishJournal for the Philosophy of Science 47(1), 1–29.

Rorty R. (1981): Philosophy and the Mirror of Nature, Princeton UniversityPress, Princeton.

224 Horst

Salmon W. (1971): Statistical Explanation and Statistical Relevance, Universityof Pittsburgh Press, Pittsburgh.

Salmon W. (1984): Scientific Explanation and the Causal Structure of theWorld, Princeton University Press, Princeton.

Schaffner K.F. (1967): Approaches to reduction. Philosophy of Science 34(2),137–147.

Schaffner K.F. (1974): The peripherality of reductionism in the development ofmolecular biology. Journal of the History of Biology 7(1), 111–139.

Searle J. R. (1992): The Rediscovery of the Mind, MIT Press, Cambridge.

Silberstein M. (2002): Reduction, emergence and explanation. In BlackwellGuide to the Philosophy of Science, ed. by P. Machamer and M. Silberstein,Blackwell, Oxford, pp. 80-107.

Sklar L. (1999): The reduction(?) of thermodynamics to statistical mechanics.Philosophical Studies 95(1-2), 187–199.

Smart J.J.C. (1959): Sensations and brain processes. Philosophical Review 68,141–156.

Stich S.P. and Laurence S. (1994): Intentionality and naturalism. In Mid-west Studies in Philosophy Vol. XIX, ed. by P.A. French, T.E. Uehling andH.K. Wettstein, University of Minnesota Press, Minneapolis, pp. 159–182.

Suppes P. (1974): The structure of theories and the analysis of data. In TheStructure of Scientific Theories, ed. by F. Suppe, University of Illinois Press,Urbana, pp. 266–283.

Toulmin S. (1974): The structure of scientific theories. In The Structure of Sci-entific Theories, ed. by F. Suppe, University of Illinois Press, Urbana, pp. 127-134.

Wilson E.O. (1998): Conciliance: The Unity of Knowledge, Knopf, New York.

Received: 12 May 2014Revised: 30 October 2014Accepted: 04 November 2014

Reviewed by David Anderson and Michael Anderson


Commentaries to Horst

Commentary by Harald Atmanspacher

The article by Horst questions the validity of reductive schemes notonly for the explanatory gap between mind and brain, but in the sci-ences in general. One can hardly state this more succinctly than this:“Reductive and eliminative physicalism are damaged, maybe without re-suscitation”. But Horst also offers inspiring thoughts about the relationbetween epistemic explanation and metaphysics, and asks us to criticallyrethink the naive “realistic inventory ontology” of most of science. Hesuggests to regard the problem of the purported disunity of science (a laCartwritght or Dupre) as “a product of how the mind conceives of theworld”. But this is not all: he also challenges us to think about howthis may or may not lead to an “ontological pluralism” that scrutinizesthe idea of a fundamental ontology of a world composed of elementarybuilding blocks.

Before I will amplify on this latter, truly momentous issue, I want toadd some comments about the “disunity of science” and what makes mesomewhat reluctant about it. First of all, it is uncontroversial that manytheoretical frameworks in the sciences appear – on the surface – disparateand perhaps even unconnected, as expressed by Dupre’s (1993) “disorderof things”. Even more critical assessments are suggested by phrases suchas Cartwright’s (1983) “how the laws of physics lie”, claiming that thefundamental laws of physics never exactly describe the behavior of objectsin the real world.

Although some of the observations by Dupre and Cartwright are notwrong, they overstate their case or are misleading. For instance, abstrac-tion and idealization are important tools in theory building not becausethey provide more or less appropriate approximations to real-world sit-uations. The truth of it is that they provide compact basic laws fromwhich descriptions of real-world situations can be exactly inferred if theirspecific context (which is not part of the basic laws) is precisely specifiedand implemented. For a number of examples in physics and chemistry,this has been successfully demonstrated using the interlevel relation ofcontextual emergence introduced by Bishop and Atmanspacher (2006) –which limited space forbids me to further expand on here.

Rather than abstraction and idealization, the key issue here is to iden-tify the contingent contexts relevant for the situation to be described.They allow us to define the properties that are relevant for this situationand connect them in a well-defined and rigorous way with the basic laws.This does not involve lying or cheating – it just takes a careful analysisrespecting all the details that the sciences provide. See Atmanspacher and

226 Horst

beim Graben (2009) for an overview of how this works and for examples,including aspects of the psychophysical gap.11

Another, different concern about unity and disunity refers to the is-sue of incompatible descriptions which physicists are familiar with fromquantum theory. This kind of incompatibility is based on a well under-stood relation between its referents: two descriptions are incompatibleif they include propositions that cannot be pasted together in a globallyBoolean fashion but form a partial Boolean lattice (Primas 2007). A maxi-mal form of such an incompatibiity is known as complementarity. Notionslike incomparability and incommensurability can be treated in the sameframework (beim Graben and Atmanspacher 2009).

All these in-...-bilities express exactly specifiable relations between de-scriptive frameworks, models or theories, recently even in cognitive sci-ence: decision making, sequence effects, etc. (cf. Wang et al. 2014). Onthe surface they seem unconnected, but underneath they tell us a lotabout the intricate ways in which connections exist and wait to be un-veiled. From this point of view, non-Boolean relations will appear asdisunities from a Boolean perspective. We may speculate whether ourcognitive architecture contains structural principles favoring such Booleanassessments. Perhaps fast yes/no decisions have been more important inevolution than the ability to sustain ambiguity and uncertainty. But evo-lution is not finished yet, is it?

I think that the disunity thesis should be seriously deflated becauseof the reasons mentioned above. However, I would like to argue that thisdoes not obliterate cognitive pluralism but rather strengthens it by itsconsequences for scientific ontology.12 This touches the option of ontolog-ical pluralism which Horst addresses, but does not seem to endorse verymuch. To indicate my own speculative outlook in this regard, a few com-ments along the lines of previous work (Atmanspacher and Kronz 1999)are appropriate. The key idea here is inspired by Quine’s (1969) onto-logical relativity, a somewhat solitary theme in his work which does notfeature crucially among those ideas that made his reputation.

The way in which we tried to refine Quine’s proposal is applicable toany domain of scientific description in which one can sensibly characterizea system by individual and statistical states and their associated proper-ties. The example of relating (classical) mechanics and thermodynamicsillustrates this in an easily intelligible way: individual mechanical statesof a many-body system with their properties (positions, momenta) can be

11I am skeptical, by the way, that the hard part of the gap, the step to phenomenalcontent, can be bridged by contextual emergence alone. This may indicate that thisgap is special indeed, a point that Horst discusses, but essentially leaves open.

12Many scientists would claim that they need no ontology anyway. But, as vonWeizsacker used to say, those who hold on to this claim most firmly are typicallyendued with an ontology of the crudest variety.


pooled in an ensemble description to form a distribution called a statis-tical state. Within a thermodynamical description, this statistical statebecomes re-defined as an individual state pertaining to the system as awhole, for instance a temperature state.

The relation between the mechanical statistical state and the individ-ual thermodynamical state is given by what Nagel called “bridge laws”(which contextual emergence allows us to systematically identify).13 Inmany other, less developed examples, most notably the relation betweenneural and mental states, we are still far from any idea of how such bridgelaws might look like – which just amounts to another formulation of thepsychophysical gap.

Ontological relativity suggests that both the individual mechanicalstates and the individual thermodynamic states can be taken to define anontology, depending on the descriptive domain to which one decides to becommitted. In the example above, classical mechanical states are usuallyconsidered as ontic relative to a thermal description. However, from aquantum theoretical point of view, classical states would be consideredepistemic, to be derived by proper superselection rules in the (Hilbert)space of ontic quantum states. And from a solid-state physics point ofview, thermal states may be considered ontic relative to the artifacts of ourdaily lives. The state of a piece of furniture will certainly be viewed onticfrom the viewpoint of a designer whose ontological commitment refers toweight, size, shape, texture etc. rather than entropy, specific heat andtemperature.

So far, these examples are sketched out within the traditional com-positional “inventory” framework. But ontological pluralism also worksin a decompositional way. In the so-called standard model of physics,the description at a most fundamental level is most symmetric, and theemergence of the fundamental forces of nature can be couched in terms ofsuccessive symmetry breakdowns. A truly fundamental “theory of every-thing” would amount to a totally symmetric description, free of decom-posing distinctions. The consequence of this complete symmetry is thatthere are no distinguishable phenomena (so a theory of everything is lit-erally a theory of nothing). Symmetries are broken by distinctions which,thus, yield phenomena – as Pierre Curie once said: “it is the dissymmetrywhich creates the phenomenon”. Which symmetries one wants to respectand which symmetry-breaking distinctions one wants to make again is amatter of ontological commitment.

This raises the question of how to ground ontological commitment.One idea which I found helpful here has been introduced, apparently in-

13Interestingly, the bridge law here, that the mean kinetic energy from the many-body momentum distribution is proportional to the temperature, has found its rigorousformal explanation not before the late 1960s – although it was heuristically known sinceBoltzmann’s work in the 19th century.

228 Horst

dependently, by van Fraassen (1980) and Garfinkel (1981): the idea ofusing relevance relations to fix the context-dependence of scientific expla-nations. On the accounts of Garfinkel and van Fraassen, explanations arenot only relationships between theories and data: they are three-placerelations between theories, data and contexts. Relevance relations aredetermined by contexts that have to be selected and are not themselvespart of a scientific explanation.14 Science tries to explain the structureof the phenomena in the world, but it does not determine which partsof that structure are salient in particular situations. To do so, one needsto set a context. This is an epistemic act entailing cognitive pluralism.It grounds the ontological commitment required in ontological relativityand, for that matter, ontological pluralism.

This farewell to the centuries-old dogma of one absolute fundamentalontology (Horst’s “naively realistic inventory ontology”) is not yet partof current mainstream thinking. But in times in which fundamentalism– in science and elsewhere – appears increasingly tenuous, ontologicalpluralism offers a viable alternative for more adequate and more balancedframeworks of thinking – and acting. The resulting picture is more subtleand more flexible than an overly bold reductive fundamentalism, and yetit is more restrictive and specific than a patchwork of unconnected modelfragments.

Atmanspacher H., Bezzola Lambert L., Folkers G. and Schubiger P.A. (2014):Relevance relations for the concept of reproducibility. Journal of the RoyalSociety Interface 11, 2013.1030.

Atmanspacher H. and beim Graben P. (2009): Contextual emergence. Schol-arpedia 4(3), 7997, accessible at www.scholarpedia.org/article/Contextual_emergence.

Atmanspacher H. and Kronz F. (1999): Relative onticity. In On Quanta, Mindand Matter, ed. by H. Atmanspacher, A. Amann, U. Muller-Herold, Kluwer,Dordrecht, pp. 273–294.

beim Graben P. and Atmanspachr H. (2009): Extending the philosophical signi-ficance of the idea of complementarity. In Recasting Reality. Wolfgang Pauli’sPhilosophical Ideas and Contemporary Science, ed. by H. Atmanspacher andH. Primas, Springer, Berlin, pp. 99–114.

Bishop R.C. and Atmanspacher H. (2006): Contextual emergence in the de-scription of properties. Foundations of Physics 36, 1753–1777.

Garfinkel A. (1981): Forms of Explanation, Yale University Pressm, New Haven.

Primas H. (2007): Non-Boolean descriptions for mind-matter problems. Mindand Matter 5, 7–44.

14More formally speaking, such relevance relations determine equivalence classesof individual states whose differences are irrelevant within the given context. Theseequivalence classes are required to form proper statistical states, a crucial issue in theconstruction of contextually emergent properties. More detailed discussion about thiscan be found in Atmanspacher et al. (2014).


Quine W.V.O. (1969): Ontological relativity. In Ontological Relativity andOther Essays, Columbia University Press, New York, pp. 26–68.

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

Wang Z., Busemeyer J.R., Atmanspacher H. and Pothos E.M. (2014): The

potential of using quantum theory to build models of cognition. Topics in

Cognitive Science 5, 672–688.

Commentary by Michael Silberstein

Horst, following the recent anti-reductionism and pro-pluralism trendsof philosophy of science, argues that successful intertheoretic reductions(at least as conceived by Positivist philosophers of science) are actuallyrare and quite spotty. Horst goes on to argue that both dualism (oranti-physicalist accounts more broadly) and physicalism are undercut byrampant pluralism, explanatory gaps and disunity in science. That is,both dualism and physicalism share the assumption made explicitly bySeager that physicalism seems like a lock for all phenomena except forthe mind. So, if Horst’s claim about the state of disunity in scienceis true then Seager’s argument that it is time to seek anti-physicalistalternatives regarding the mind does not go through. And obviously,if Horst is right about the state of science then physicalism is no morejustified than dualism or other brands of anti-physicalism.

Horst argues that the disunity in science is really an artifact of the in-herent “Cognitive Pluralism” of the human mind. On this view it wouldseem that Kantian-like categories are still with us, shaping all our experi-ence, providing access only to the phenomenal world while the noumenalremains forever beyond our reach. This is in many respects a deflationarymove when it comes to the metaphysical project of analytic philosophy:

To this I have added an outline of what I take to be a promisingway of proceeding: by re-examining philosophy of mind from theCognitive Pluralist standpoint, which holds theory pluralism to bea product of how the mind conceives of the world, and to explore thecognitivist turn in metaphysics as an alternative to naively realisticinventory ontology.

If, however, we are concerned with the metaphysical sense of “nat-uralism”, we must separate Cognitive Pluralist epistemology fromCognitive Pluralist metaphysics. The epistemology is one that,as I have indicated, some naturalists (particularly non-reductivephysicalists) might well find initially attractive. As a metaphysi-cal position, however, Cognitive Pluralism would seem to be anti-naturalistic in much the same ways as are transcendental idealismor pragmatism. These are all positions that reject the assumptionthat, when we have found the most fundamental posits of natural

230 Horst

science, we have reached philosophical bedrock as well: there is afurther, and indeed a deeper, story to be told about what it is tobe an object, even an object of a fundamental physical kind.

Horst’s position suggests abandoning the metaphysical project in favorof the exploration of Cognitive Pluralism and is therefore best seen as aninstance of what Seager, in his paper in this issue, characterizes as arejection of ultra-strong scientific realism:

Finally, one can modify one’s metaphysics. In some ways, this isthe most radical reaction to the problem of consciousness. Crudelyspeaking, this option requires one to give up a kind of ultra-strongscientific realism, or as it might be more pejoratively called, sci-entism which is not unfamiliar: the view that science is the solesource of knowledge about the ultimate nature of reality.

This rejection of ultra-strong scientific realism is something Horst shareswith many other thinkers such as Dupre, Cartwright and van Frassen, buthis reasons for this rejection are quite distinct from others. For example,he rejects the radical (metaphysical) pluralism of Dupre and Cartwright,and he does not reject methodological or epistemological naturalism infull (Horst 2007). Horst’s claim, given Cognitive Pluralism, is that we areepistemically in no position to make grand metaphysical claims whatso-ever.

Given that Horst’s rejection of ultra-strong scientific realism is basedon his belief in Cognitive Pluralism, which in turn is based on his claimsabout the disunity of science, let me note that this is the first difference Ihave with him. Namely, I think science is more unified (or at least moreopen to unification) than Horst gives it credit for, and therefore I do notthink Cognitive Pluralism is as well motivated as Horst would like. Toparaphrase myself from a review of Horst’s 2007 book, Beyond Reduction(Silberstein 2011, p. 571), what Horst has shown is not that intertheoreticreduction is rare or impossible, but rather that the deductive-nomological(normative) model of the Positivists is a bad one.

To be clear, however, I am more in agreement with Horst than Seagerabout the state of disunity in science, and as Horst notes I have argued asmuch elsewhere (Silberstein 2002). I would say that the truth about sci-ence lies somewhere between Seager’s neo-unified-science-hypothesis andHorst’s pluralism. However, while I am not persuaded by Horst’s ar-gument from Cognitive Pluralism against ultra-strong scientific realism,I also reject this doctrine if, as Seager claims, it is co-extensional withscientism.

In the same sense in which “contextual emergence” is driven by andcontinuous with science, I believe scientific realism is a case-by-case af-fair. That is to say, while my arguments for contextual emergence andneutral monism are driven by science, I don’t believe science is the only


source of knowledge. I am struck by how often people portray scienceas a monolithic, homogenous enterprise with its own shared, given andfixed worldview. As if to embrace science or moderate scientific realism isto embrace physicalism, the causal closure of the physical, or scientism.From this perspective, if you want to avoid one or more of these doctrinesyou had better find a way to dethrone science or take it down a peg ortwo.

I embrace science as the primary basis for doing metaphysics aboutthe natural world, but I reject the three aforementioned doctrines. Howis this possible? As I argued in my main paper, the best interpretation ofscientific evidence across the board tells in favor of contextual emergenceand therefore against these three dogmas. I am arguing from science,not against it. But there is nothing in contextual emergence that entailsultra-strong scientific realism/scientism.

Seager and I disagree about how best to interpret what science is tellingus about the world. Seager argues that science is telling us that physical-ism (or at least reductionism in the actual world) is true except for themental domain. I am arguing for a contextual-emergence interpretation ofscience. Horst, on the other hand, argues for a Cognitive Pluralism inter-pretation of science. From my perspective the situation looks like this: tothe extent allowable by real-world scientific underdetermination, Seagerand I want to have a straight contest between physicalism/reductionismand contextual emergence; may the best view win. Horst wants to callthe game before it is even over, because it is simply too dark to see orotherwise know who is winning. Of course, Seager and I might be wrongand Horst right, but from my perspective the game should go on muchlonger before it is called due to intrinsic and insurmountable ignorance.This is why in my review of Horst’s book I called Cognitive Pluralism a“God-ofthe-gaps” type move.

I also worry that if the argument from Cognitive Pluralism cuts againsteven moderate scientific realism, then it has a similar problem to globalskepticism. Namely, that in undercutting our basis for knowledge, we alsoundercut any reason to believe that global skepticism is true. For example,Horst’s argument for Cognitive Pluralism is based partially in cognitivescience (i.e., naturalized epistemology and evolutionary psychology), inparticular in a modularistic conception of mind and cognition. I worry thefollowing position is self-defeating for obvious reasons: “cognitive sciencetells us Cognitive Pluralism is universally true”. In short, given CognitivePluralism, how can we put all of our faith in any overarching paradigmof cognition?

But putting all this aside, I have argued in my main paper and else-where that we should reject modularity, the computational theory of mind,the representational theory of mind, and the localization/decompositionof brain mechanisms (Silberstein 2006, Chemero and Silberstein 2008, Sil-

232 Horst

berstein and Chemero 2013). We should reject these in favor of embodied,embedded and extended cognitive science. This was discussed in my mainpaper, where I also argued that extended cognitive science is an exam-ple of contextual emergence. For other excellent and cogent criticisms ofcomputationalism and representationalism, I suggest Horst read Symbols,Computation and Intentionality: A Critique of the Computational Theoryof Mind (Horst 1996).

This brings me to my second disagreement with Horst. From myperspective, the better explanation for the failure of unification is notCognitive Pluralism, but contextual emergence. Science to be sure isoften characterized by explanatory pluralism, computational complexity,the idealization of scientific models, diverse explanatory interests, differ-ent characteristic properties and accompanying vocabularies at differentspatiotemporal scales and energy levels, in different contexts, at differentlevels of complexity, etc. Horst is right to emphasize these aspects of sci-entific practice. His description of scientific models is not the problem, asexplanatory pluralism is indeed a fact of life in the sciences.

But explanatory pluralism does not entail Cognitive Pluralism, thatwe are hardwired in a neo-Kantian fashion for conceptual and theoreticaldisunity. Giving up on Positivist dreams of the unity of science, giving upon the idea of axiomatizing scientific theories, etc., does not mean givingup on unification altogether in any form. Explanatory pluralism and unityreasonably conceived are not mutually exclusive if one takes contextualemergence seriously. And again, as I said in my main paper, contextualemergence allows for conceptions of scientific unity that are not verticalor reductionist, but rather horizontal. With contextual emergence thereis plenty of unity/monism and disunity/pluralism to go around.

So what I want most is to persuade Horst that contextual emergencecan explain scientific pluralism without giving up moderate scientific real-ism or naturalism. Neutral monism as contextual emergence is a straight-forward way to win the game – an interpretation of science that showsscience itself is the best evidence against ontological and explanatory re-ductionism.

Chemero A. and Silberstein M. (2008): After the philosophy of mind: Fromscholasticism to science. Philosophy of Science 75, 1–27.

Horst S. (1996): Symbols, Computation and Intentionality: A Critique of theComputational Theory of Mind, University of California Press, Berkeley.

Horst S. (2007): Beyond Reduction: Philosophy of Mind and Post-ReductionistPhilosophy of Science, Oxford University Press, Oxford.

Silberstein M. (2002): Reduction, emergence, and explanation. In BlackwellGuide to the Philosophy of Science, ed. by P. Machamer and M. Silberstein,Blackwell, Malden, pp. 80–107.


Silberstein M. (2006): In defense of ontological emergence and mental causation.In The Re-emergence of Emergence, ed. by P. Clayton and P.S. Davies, OxfordUniversity Press, Oxford, pp. 203–226.

Silberstein M. (2011): Metaphysics or science: The battle for the soul of phi-losophy of mind. Philosophical Psychology 24, 561–573.

Silberstein M. and Chemero A. (2013): Constraints on localization and de-

composition as explanatory strategies in the biological sciences. Philosophy of

Science 80, 958–970.

Commentary by William Seager:Cognitive Pluralism and Ontological Bedrock

Although (I promise) Horst’s paper and mine were composed in com-plete independence, they show some interesting points of convergence anda kind of broad agreement about the range of options available when wespeculate about the nature and future of physicalism. But since discussingpoints of agreement would likely be rather dull, and violates venerablephilosophical norms, I will focus on points of disagreement or aspects ofHorst’s paper which I find unclear or incomplete.

One of the latter concerns involves Cognitive Pluralism and how itwould reformulate our understanding of nature as compared with the sortof understanding standard physicalism aims to attain. Cognitive Plural-ism is set in opposition to both reductive physicalism and realist pluralism.A standard division within the ambit of reductive physicalism (which Iappealed to in my paper but which is very common currency) is thatbetween epistemological and ontological reductionism.

Epistemological reductionism is a kind of ideal or limiting case of uni-fication, in which all theoretical concepts (ultimately, all concepts of anykind) take their place in a hierarchical structure in which their patterns ofinstantiation are a priori consequences of a set of (instantiations of) priv-ileged, fundamental concepts (including characterizations of both entitiesand laws which govern their evolution and interactions). I add the guffabout “patterns of instantiation” simply to forestall the obvious objectionthat no deductive argument will introduce syntactically new concepts ex-cept in trivial ways. The picture is that the patterns and entities ofnature to which non-fundamental concepts apply will be deducible fromthe patterns of instantiation of the fundamental concepts.

It is perfectly legitimate to allow the reductive standpoint to possesshigher-level concepts, whose patterns of instantiation will be found tobe entailed by the arrangement of fundamental entities given the lawsgoverning those entities. Fundamentality will be defined in terms of theserelations of entailment. The fundamental concepts are, for the physicalist,those of our most basic physical theory (plus those concepts which form

234 Horst

the indispensable machinery of mathematics and logic which are, for thephysicalist, harmless because content-free, although of course this is anadditional highly contentious philosophical claim which the physicalistmust defend).

This view was succinctly expressed by Einstein (Einstein and Seelig1954, p. 226):

The supreme test of the physicist is to arrive at those universal lawsof nature from which the cosmos can be built up by pure deduction.

Pedantry insists we add that some initial conditions will be needed if weare to deduce our universe as opposed to merely the set of nomologi-cally possible universes. Note that there are two features to Einstein’sdream: the a priori construction and the presumption that physics pro-vides the basic and complete construction material. One might denyphysicalism but accept the idea of a priori constructionism. One suchanti-physicalist, at the opposite end of the succinctness spectrum fromEinstein, is Chalmers (2012; following on from Jackson 1998 and Carnap1928/1968).

If we step back from physicalism, does some form of epistemologicalreductionism turn out to be clearly true? Suppose we know everythingabout our world except for where the Barber Shops are. Could we then apriori deduce their locations from our knowledge base? It certainly seemsso. So epistemological reductionism about Barber Shops – at least theirlocations – relative to that exceptionally comprehensive knowledge baseappears to be true. The question is of course how much can we whit-tle down the knowledge base of “fundamental concepts” while retainingcomplete deducibility?

Realist pluralism is thus not in opposition to a priori deducibility,for surely not every feature or aspect of the world is irreducible. Realistpluralism only makes sense in opposition to a specific proposed set offundamental concepts. For example, Chalmers thinks that phenomenalityneeds to be added to the catalogue of the fundamental, but, for him,pretty much everything else follows the dictates of standard reductivephysicalism. Horst notes that the realist pluralist denies that

the phenomena of [the] natural sciences are metaphysically deter-mined by the complete set of physical facts and laws cast at thelevel of the simplest physical objects.

Such denial leaves entirely open that such determination is to be found,for example, if we consider basic physical kinds plus chemical kinds. Thatwould be a view that embraced some kind of robust chemical emergence,but from a philosophical perspective it would be only notionally differentfrom standard reductionism. It thus seems that the difference betweenrealist pluralism and physicalist reductionism is only one of degree: a


parochial debate about the size of the fundamental base. Of course, ifthe base gets really really big, the result won’t look much like physicalistreductionism, but Horst himself appears to agree that, from within theshared strictures of the opposition between realist pluralism and reduc-tionism, the fundamental base is likely to be fairly small.

I think it is Horst’s Cognitive Pluralism that stands out as radicallynovel. There are two aspects I want to focus on. The first is the at-tack on possible-worlds semantics. Horst characterizes possible worlds inpropositional terms:

Consider how one gets at the notion of a possible world. One startswith a set of all propositions or a set of all states of affairs. A worldis then conceived of as being, or as being defined by, a mapping fromthis set onto truth values. A possible world is one in which thereare no inconsistencies generated by such a mapping.

One can look at possible worlds quite differently. One can view the possi-ble worlds as constituting modal space. Possible worlds are the fundamen-tal entities here, not derivatively dependent on our powers of descriptionor conceptualization (our access to them is perhaps so constrained). Ifone takes the idea of a possible world as basic, propositions are defined interms of (sets of) possible worlds rather than vice versa.

Do we have a notion of the world? The world is everything that exists.So, yes, we have an idea of the world. Things could have been different,either in what exists or their arrangements. So we have a notion of possibleworlds. Although obviously compatible with it, this does not entail aLewis-style realism about possible worlds. It is simply a consideration infavor of the idea that we do have a notion of the world as the totalityof existence and hence have an idea of possible worlds independent ofparticular modes of describing them.

This way of thinking about possible worlds does not demand that therebe “some unique, canonical, mind-independent way that reality dividesitself up into properties, propositions or states of affairs”. The possibleworlds are what they are, and they support description in terms of allpossible “divisions” into properties, etc. Some worlds make particulardescriptions come out true, others make them come out false. If there areincommensurate conceptual schemes (schemes which neither contradictnor agree with each other’s nontrivial descriptions), possible worlds haveno trouble accommodating them. A true proposition is defined as onewhich includes the actual world as an element. It is impossible for aproposition and its negation to both be true because of elementary settheory. Consistency of truth falls out naturally, and we can all join theright-thinking ranks of those who hold that truth entails consistency.

However, this says nothing about truth conceived of as an “honorific”or as a symbol for the usefulness of some theory. It is natural to conceive of

236 Horst

this kind of “truth” in terms of sets of possible worlds in which certain keyvalues of interest to us fall within certain ranges. For example, Newtoniangravitation is honorifically true because the gravitational force betweentwo bodies which it predicts is within some distance of the true force,where this distance is small enough to ignore for our predictive purposes.The set of worlds that define this range is, presumably, pragmaticallydetermined. The proposition expressed by this set of worlds would besomething like: Newtonian gravitational theory predicts the force betweenbodies in configuration X to within Z newtons. If the actual world isin this set then the proposition is true and Newtonian theory is indeedaccurate in these circumstances within the specified range of accuracy.Hence, Newtonian theory is honorifically true for some purposes but notfor others (if non-Newtonian factors were neglected, our GPS systemswould be worthless).

Nothing in this possible-worlds semantic framework requires that therebe just one overarching theory or description of the actual world that isfully accurate, even as it cleaves to the twin ideas that truth impliesconsistency and that the world is exactly one of the ways it could be.

Despite Horst’s antipathy to possible-worlds semantics, I don’t seeanything in the foregoing which is opposed to Cognitive Pluralism. Infact, I think it fits perfectly with the second striking feature which Horstadvances in his characterization of Cognitive Pluralism. I think this isthe most exciting feature of the view, which I will venture to charac-terize as (maybe Horst would not like this) the mind-dependence of theworld. Of course, I don’t mean that without minds nothing would exist(we could always say, leaving aside the issue of modal realism, that thepossibilities would exist without minds15). I take it that the sort of mind-dependence endorsed by Cognitive Pluralism arises from the requirementthat thoughts be conceptually structured along with the denial that thereare privileged, so to say “really real” conceptualizations that uniquelycapture the genuine metaphysical structure of the world.

Horst puts it thus:

what it is to be an object is to be the sort of thing that can bemade a possible object of cognition. Objecthood is understood interms that relate it to thinking subjects.

Leaving aside much complication we can extend this point beyond objectsto the “lived world‘” which could then be described as (van Fraassen 1980,p. 81) “the intentional correlate of the conceptual framework throughwhich [we] perceive and conceive the world”.

15If we are modal realists, on the other hand, we can quickly and decisively an-swer Leibniz’s old question of why there is something rather than nothing. Supposingnothing to exist, it would remain possible for something to exist.


The conceptual constraints that underlie quotidian claims about andapprehensions of objects, properties, causal relations, etc. are doubtlessvery complex. But I wonder what Horst would think of applying the sim-ple formula of Dennett (1991), that the core constraint is efficient andeffective “packaging” of patterns. On Dennett’s view (as I understand it),an object counts as real if its inclusion in our conceptual scheme providesan efficient way to summarize a range of possible experiences, measure-ments or observations (so it is closely related to the pragmaticist ratio-nale). Although Dennett might resist any attempt to interpret his viewsas deeply metaphysical, his discussion of real patterns seems congenial tothe cognitive pluralist.

Yet patterns need material. Although it is fair to say that we are “al-ways already” in our world, which generates material enough to define ordiscover patterns in “it”, this “material” is conceptually structured andwhatever that means exactly, it is an activity of consciousness. We fre-quently appeal to Nagel’s “what it is like” rubric but we could, I think,equally characterize consciousness as there being a world “for the subject”(although there is no claim that the subject consciously apprehends thenature of its world and its place within it – only self-conscious beings canattain to this). Only conscious beings find themselves in a world in thestrong sense of being conscious of its objects and structures. The worldthey find themselves in is not some bare thing-in-itself, but a world struc-tured by their conceptual (perhaps more broadly, their representational)resources. The pattern metaphysics operates over this domain.

But that leaves the domain of “patterning”; the domain of conscious-ness itself which cannot be reduced to just another “real pattern”. Thisis perhaps another way Cognitive Pluralism has to take consciousness asspecial. It is a power of “world creation” and cannot be regarded assimply another possible aspect of worlds. In this sense, consciousness isontological bedrock.

Carnap R. (1928/1968): The Logical Structure of the World and Pseudoproblemsin Philosophy, translated by R. George, Routledge and Kegan Paul, London.

Chalmers D. (2012): Constructing the World, Oxford University Press, Oxford.

Dennett D.C. (1991): Real patterns. Journal of Philosophy 88(1), 27–51.

Einstein A. and Seelig C., eds. (1954): Ideas and Opinions, tanslated by C. Barg-mann, Crown, New York.

Jackson F. (1998): From Metaphysics to Ethics, Oxford University Press, Ox-ford.

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

238 Horst

Replies by Horst

For a philosopher, it’s already a good day when three respondentsagree that you have raised serious problems for all of the mainstreamviews in philosophy of mind and articulated an interesting and engagingphilosophical alternative. When each of them raises interesting points ofengagement, the day is getting even better. Some of the issues raisedwere new to me, and others topics that I had already begun to thinkabout in the months since the conference, in working to complete a bookon Cognitive Pluralism. As each respondent addressed a unique set oftopics, I shall divide my response by commentator.

Reply to Silberstein’s Comments

My paper in this journal issue, like my book Beyond Reduction (Horst2007), can be broken into two independent parts: (1) an argument thatthe failures of inter-theoretic reduction in science should cause us to re-think the whole mainstream conversation in metaphysics of mind, and (2)a presentation of Cognitive Pluralism (CP) – a view on which the cogni-tive architecture of human understanding is based in idealized models ofvarious content domains – along with a brief account of how CP wouldmake various sorts of scientific disunity less puzzling, and a sketch of itspossible larger implications for epistemology and metaphysics.

Silberstein’s response includes two interpretations of the relationshipbetween these parts. One of these, which is the correct interpretation, isthat CP is primarily a theory of cognitive architecture, whose credentialsultimately have little to do with questions of scientific unity, beyond thefact that it has epistemological implications that make disunity unsur-prising (although CP is also compatible with the possibility that there isa way of bootstrapping past our architectural bias towards disunity).

The other interpretation, which is suggested by the fact that (1)and (2) are presented together in the fashion I have presented them,is that I am arguing for CP (or, more carefully, first for CP as a cog-nitive/epistemological thesis and then for a cognitivist, pluralist meta-physics) on the basis of scientific disunity. While I do think that CP’sability to explain scientific disunity is evidence in its favor vis-a-vis theo-ries that are incompatible with or vexed by disunity, I do not think that anexplanation of scientific disunity could serve as an adequate justificationfor CP or any other theory of cognition.

In one sense, CP is introduced as a way of re-orienting our views ofscience and the mind as an alternative to the mainstream views that arethreatened by scientific disunity. In another sense, the critique of contem-porary philosophy of mind is a prelude to the introduction of an alternative


view of understanding, epistemology, semantics, and metaphysics that Ithink is ultimately more interesting than the problematic it is supposedto replace.16

It is beyond the scope of this discussion to fully compare CP to com-petitor theories of cognition. In my forthcoming book on Cognitive Plu-ralism, I frame the main point of contention as one between model-basedviews of understanding and views that treat beliefs or propositions as theprincipal units of understanding. As I say in my comments on Silber-stein’s article, though, CP has an interesting self-applicability: Because itis itself a model of particular aspects of the mind, we should not expectit to be the only important story about the mind, and should expect thatsome alternatives may be more complementary than competitors.

I have gradually come to regard ecological/enactive accounts like Sil-berstein’s in this light. To the extent that CP is a “cognitivist” theorywhich brackets interactions with the world to attend to how we “repre-sent” the world, it needs complementary accounts that are framed withbroader scope, both to deal with interactive online cognition and to ad-dress broad notions of content. Conversely, while the dynamical systemsframework may in principle be able to accommodate both online and of-fline cognition, it ultimately needs some kind of story about what kinds ofsystems within the mind are used in offline cognition (and perhaps also foronline cognition that also involves reasoning), such as the view that thisinvolves mental models that mirror external affordances, to which otherinternal processes are coupled in lieu of the affordance themselves.

Silberstein is correct in calling for further and more detailed engage-ment on the questions of how much and what kind of unity is to be found inthe sciences. My paper was framed around a critique of a particular influ-ential notion of scientific unity – reductive unity – though it also toucheson another important type of unity/disunity: consistency/inconsistencyof theories and the inferences they license. Philosophers have been drawnto reductive unity in part because it yields metaphysical supervenience,and the standard of consistency is generally regarded as an inviolable log-ical norm; and so, if we have to do without these, that is a pretty big dealfor theories of metaphysics, mind, logic and truth. But from the stand-point of philosophy of science, the real work of mapping the relationshipswe do find between models is just beginning.

I thus by no means regard Cognitive Pluralism as a view that im-plies that we should give up on looking for connections or for more globalframeworks. Indeed, I think that the search for such things is a kind ofregulative ideal of reasoning. My objection is to assuming that it can ac-

16It is not necessarily more important though, as issues about reduction have longbeen perceived as a threat to our humanistic self-image. To my mind, the most impor-tant take-home message from Beyond Reduction is that such anxieties are unnecessary,because the kind of reductionism that incited them has not panned out.

240 Horst

tually be accomplished, and particularly that some one vision of how it isto be accomplished (in the last century, the vision of reductionism) shouldbe treated as obviously correct. And one thing that I think Cognitive Plu-ralism adds to this general admonition is the idea that the challenges towould-be comprehensive views come not only from recalcitrant data, butalso from insights that we seem to only be able to obtain from differentcognitive perspectives.

Reply to Atmanspacher’s Comments

Atmanspacher agrees with me that we should bid adieu to the ideaof a single, canonical, mind-independent inventory ontology, and indeedsupplies additional arguments for this conclusion that were new to me,which I intend to add to my bag of tricks in the future. I am instructedand edified by Atmanspacher’s observations on symmetry breaking andrelevance, and by his conclusion that

... a truly fundamental “theory of everything” would amount toa totally symmetric description, free of decomposing distinctions.The consequence of this complete symmetry is that there are nodistinguishable phenomena so a theory of everything is literally atheory of nothing.

This does indeed seem to imply the need for a pragmatically-driven theorypluralism. And if one thinks that relevance requires a cognitive act ofsetting a context, it implies a cognitive form of pluralism, and on groundsthat do not depend upon a commitment to a particular form of cognitivearchitecture.

Atmanspacher also agrees that there are numerous idealized laws,models, and theories that carve up the world in different ways. But heargues that I have overstated the implications this has for the disunity ofthe sciences. For example, complementary wave and particle descriptions,in spite of their incommensurability, do not lead to inconsistencies or di-vergent predictions because they are applicable only to mutually-exclusivecontexts. I have found myself confronting this sort of issue in completinga book on Cognitive Pluralism subsequent to the conference from whichthis paper emerged, and indeed think now that the pragmatic characterof models, especially their being aptly applied only in particular contexts,often offsets potential incompatibilities that might otherwise result fromtheir representational differences.

Of course, there are also cases where combining the best theories wepossess yields truly problematic results, such as the application of relativ-ity and quantum theory to very dense matter, or their different assump-tions about local causation and frame invariance. The different implica-tions of theories of species based in cladistics and reproductive isolationseem to raise different kinds of issues, more akin to the different verdicts


consequentialist and deontological ethical theories can render about thegoodness of actions.

I think that it would be a worthy undertaking to try to classify thedifferent types of situations we encounter involving dissonance betweenmodels. It may be that complementarity is an unusual situation, or thatit is, upon careful review, the norm. Or similarly for the apparently incom-patible assumptions and implications of relativity and quantum theory, orthe alternative understandings of species, which seem to capture differentbiologically salient attributes without much prospect for an experimentumcrucis to arbitrate between them.

The examples in the previous paragraph involve two kinds of cases:multiple models of what we assume to be in some sense “the same phe-nomenon” (e.g., light or species) and models of different phenomena thatdo not seem to play well together (e.g., gravitation and strong/weak/elec-tromagnetic forces). Atmanspacher also raises what I take to be a separateset of concerns about idealization when he writes that

abstraction and idealization are important tools in theory buildingnot because they provide more or less appropriate approximationsto real-world situations. The truth of it is that they provide com-pact basic laws from which descriptions of real-world situationscan be exactly inferred if their specific context (which is not partof the basic laws) is precisely specified and implemented. For anumber of examples in physics and chemistry, this has been suc-cessfully demonstrated using the interlevel relation of contextualemergence introduced by Bishop and Atmanspacher (2006) whichlimited space forbids me to further expand on here. Rather thanabstraction and idealization, the key issue here is to identify thecontingent contexts relevant for the situation to be described. Theyallow us to define the properties that are relevant for this situationand connect them in a well-defined and rigorous way with the basiclaws.

Given the content of the article referenced (which explores the relation-ships between molecular structure and chirality and between thermal equi-librium and temperature), I take it that the point here is about the rela-tionships between models at different levels of explanation. I think we areactually in substantial agreement that applying a model appropriately toa given case requires more than a knowledge of the content of the model– it requires an understanding of what cases it can be applied to, howit is to be applied, and hence specific knowledge about contexts. In myterminology, this is all part of how the content of models is “idealized”,though my view is that the reason for idealization lies in the kind of prob-lems Herb Simon outlined long ago about getting a finite mind to gainepistemic traction upon an indefinitely complex world.

But I would also stress that, when the issue is not the relations between

242 Horst

models at different levels but the relation between models and real-worldbehavior, one of the chief issues is that an important part of the contextis what other causal factors are also at work in the situation, in additionto those addressed by the model we are using to understand it. If by“descriptions of real-world situations” we mean predictions of real-worldbehavior, I doubt that this can be done exactly except in the limitingcases where other causal contributions have been excluded by the kind ofexperimental setup that Cartwright calls a “nomological machine”. (Thiswas one of the consequences of bracketing idealization I explored in Laws,Mind, and Free Will (Horst 2011).)

Reply to Seager’s Comments

The discussions of the metaphysical implications of Cognitive Plural-ism (both in the article and in Horst 2007) to which Seager responds wereadmittedly sketchy and programmatic. By this I do not mean that therewas, at the time I wrote them, a more thoroughly worked-out argumentof which I was presenting a mere sketch, but that they were initial ex-plorations of ideas that would eventually need to be worked out morecarefully at a later date. Seager’s commentary has hit upon several ofthe complications that I found myself facing in recent months in trying todevelop them more carefully.

One of these begins with the observation that the way I characterizedpossible worlds semantics is not the only way of doing so, and that notall share a commitment to what I describe as “naive realism” – the viewthat the world divides itself in a single, canonical, and mind-independentway into objects, kinds, properties, relations, and processes. Indeed, eventhe way I developed it, in terms of “the set of all propositions”, might becompatible with a cognitivist and pluralist view of concepts and proposi-tions. There would presumably be a lot more concepts and propositionsthan the naive realist might have supposed, and they would be cashedout in terms of ways that thinking beings might conceptualize states ofaffairs, but we are dealing with an infinite number of possible states ofaffairs in any case, so mere quantity should not be an issue. The issue, as Inow see it, has to do with (a) consistency and (b) the suitability of meta-physical intuitions for hitting upon metaphysical truths about necessityand possibility.

Let’s take two views of the mind: Descartes’ idea that it is an imma-terial thinking substance and Aristotle’s hylomorphic view on which thesoul is the form of the body. Assume (perhaps tendentiously) that bothDescartes and Aristotle had internally coherent models, each involving aconcept of mind or soul, which we may designate CD and CA, respectively.The concepts have different constitutive implications and hence give riseto different intuitions about what is necessary: CD entails disembodi-ment, CA embodiment. Of course, a metaphysics of necessity is directly


concerned with properties rather than concepts, and so either the twoconcepts express the same property or they express different properties.

If CD and CA are concepts of the same property, then either (1) theintuitive entailments are both metaphysically necessary for that property,but jointly generate a contradiction or (2) at least one of them misrep-resents the property it expresses, in which case at least one set of meta-physical intuitions is undependable. But we might reasonably decide thatwhich properties concepts express are differentiated by their constitutiveinference patterns, in which case there are two properties, PD and PA.It is necessarily the case that anything that is a PD is disembodied, andthat anything that is a PA is embodied, and hence nothing can be botha PD and a PA.17

Some people have intuitive senses that only the Cartesian concept isaptly applied to us, others that only the Aristotelian concept is, and manyof us feel the force of both intuitions depending on how we are primed.The longevity of these conflicting intuitions suggests that depending onsuch intuitions is not a trustworthy guide in metaphysical reasoning. Ofcourse, even if we had only one concept of mind, it might still misrepresentits target property, in which case any metaphysical intuitions based in thatconcept would be more on the order of a forced error.

The fact that there are multiple and conflicting concepts and models,each of which exerts considerable intuitive force, just brings home moreclearly the potentially misleading character of metaphysical intuitions.The sense of metaphysical necessity is a consequence of the inferentialnecessity involved in the constitutive rules of the model, and hence suchintuitions can be no better than the models they are based in. Whileconflicting metaphysical intuitions are puzzling, they may in a sense be ablessing, as they alert us to ways in which intuition-based metaphysicalreasoning may be an enterprise that goes beyond what minds like oursare suited to doing.

Seager also correctly anticipates another direction in which I have beendrawn in further developing Cognitive Pluralism: the need to characterizemodels as aiming to track “real patterns”. It is, however, quite difficult touse this idea more than heuristically without falling either into a kind ofrobust realism (perhaps of a quasi-Platonistic variety) about patterns orelse leaving it as a claim about empirically robust patterns of experience,which risks saying nothing about the aptness of that experience. I thinkboth of these options are to be avoided, but I am not sure I can do betterthan Dennett’s remarks about the duality between patterns and patternrecognition (Dennett 1991, p. 32):

17Whether either property has real instances is of course left unsettled, though theycannot be co-instantiated on pain of inconsistency and hence, if only consistent statesof affairs make for possible worlds, of impossibility.

244 Horst

in the root case a pattern is “by definition” a candidate for patternrecognition. (It is this loose but unbreakable link to observers orperspectives, of course, that makes “pattern” an attractive term tosomeone perched between instrumentalism and industrial-strengthrealism.)

Seager concludes his response with the following remarks:

But that leaves the domain of “patterning”; the domain of con-sciousness itself which cannot be reduced to just another “real pat-tern”. This is perhaps another way Cognitive Pluralism has to takeconsciousness as special. It is a power of “world creation” and can-not be regarded as simply another possible aspect of worlds. In thissense, consciousness is ontological bedrock.

I am actually largely in agreement with this. If “things” and “theworld” are to be understood as things-as-understood-through-concepts,“thinghood” is not truly fundamental in its own right, and any commit-ment to an ontological inventory of things must also involve a commitmentto at least one thinking subject as well. But this is a tricky point, and eas-ily misunderstood, as evidenced by the continuing string of philosopherswho object to Kant’s similar claims because they take them to imply thatnothing existed before thinking beings existed. The priority of consciousintentionality over world of objects is a transcendental, rather than acausal or temporal priority.

Bishop R.C. and Atmanspacher H. (2006): Contextual emergence in the de-scription of properties. Foundations of Physics 36, 1753–1777.

Dennett D.C. (1991): Real patterns. Journal of Philosophy 88(1), 27–51.

Horst S. (2007): Beyond Reduction: Philosophy of Mind and Post-ReductionistPhilosophy of Science, Oxford University Press, New York.

Horst S. (2011): Laws, Mind, and Free Will, MIT Press, Cambridge.

Beyond Reduction: From Naturalism to Cognitive Pluralism

Documents

Transcript of Beyond Reduction: From Naturalism to Cognitive Pluralism