CC BY
34
ЭПИСТЕМОЛОГИЯ & ФИЛОСОФИЯ НАУКИ • 2014 • Т. XLI • № 3
I
COMPLETE, BUT REAL. A CONSTRUCTIVIST ACCOUNT
OF REFERENCE
Tian Yu Cao -
Department of Philosophy, Boston University. E-mail: tycao@bu.edu.
1
Various theories of referent are critically but briefly surveyed from the perspective of structural realism; a constructivist version of structural realist account of referent is outlined, and its implications for history of science and for descriptive metaphysics are briefly indicated.
Key words: reference, realism, constructivism, history of science, metaphysics.
I. Introduction
Structural realism is a position which, in the last few decades, has been discussed mainly by historians and philosophers of science. In my own case, and for many others, the material is taken almost exclusively from fundamental theoretical physics, quantum physics in particular (Cao, 2010). In this paper, however, I wish to widen its scope and somewhat shift its perspective, thereby to facilitate the desirable dialogue between those from history and philosophy of physics and those with interest in the philosophical issues debated in fundamental theoretical sciences other than physics, in biology (evolutionary theory and genetics) and neuroscience, in logic, language and cognition, in philosophy of mind, and in understanding reference and realism from more general perspectives.
In any mature fundamental theoretical science, there is a fundamental ontology or a set of them, whose presence and activities are the ultimate resource the theory can utilize for describing, explaining and predicting empirical phenomena. The fundamental ontology can take various categories, such as objects or entities (an extension of objects to non-object physical entities, such as fields or physical structures), properties and relations, events and processes. But since there is no bare entity without any property, without being involved in relations with other entities, in events and processes, and there is no ontological category that has a free floating existence without being anchored in some physical entity, traditionally, more often than not, physics assumes some physical entity as its fundamental ontology, although property, process and other non-entity ontology were also occasionally suggested, such as energy in energetics and process in S-matrix theory.
Since the empirical content of a theoretical science is organized around fundamental ontology, which has some assumed properties and relations with the derivative entities that are usually suggested by observations, the existential status
1 A talk delivered at Moscow Workshop on Scientific Realism, November 22-23,2012.
72 Epistemology and Cognition
of the fundamental ontology is crucial for scientific realism. Put in the language of the Ramsey sentence formalism, should we take the fundamental ontology, a theoretical term for sure, only as a way of organizing the observable content, or as having existence in the real world? For realists, any fundamental ontology, as a natural kind term, must refer to kind of things that are to be found in nature, and the "kind" itself is constituted and individuated by some underlying factors existing in the world.
In this regard, scientific realism was seriously challenged by Thomas Kuhn with his theory of scientific revolutions. If fundamental ontology of physics, which is assumed to be the fundamental ontology of the physical world, has undergone radical changes over centuries, from Aristotle's natural places in his finite cosmos to Newton's forces in his infinite universe, to Einstein's cosmology dictated by his gravitational fields, with each incommensurable with others, Kuhn contested, then how could we take any of them seriously or realistically? Even more serious is the prospect that since there is no end to scientific revolutions in the future, no fundamental ontology can survive the radical change, and thus no future fundamental ontology can have better chance than those in the past. If fundamental ontology has no referent in the world, then scientific realism collapses.
Realists fight back by arguing that successful scientific theories must have something to do with reality, otherwise their successes could only be taken as miracles. But having relevance to reality is far too weak an argument for the reality of fundamental ontology. First, the kind term may not have any referent in the world. Caloric theory enjoyed many empirical successes, and thus must have something to do with reality, but there is no caloric as a kind of thing existing in the world. More interesting is the cases in which some theoretical terms carry reliable information about the real world, and thus can play important roles in the « theoretical structure and in connecting theory with empirical phenomena, "g but still, don't exist as a kind of entities as the theory ascribed them to be. S) Readily available examples in theoretical physics are numerous. Bare y particles with bare mass and bare charge before the renormalization ^ procedure may belong to this category (Cao and Schweber, 1993), or jg perhaps not if we can find a way to remove all other things so that as ^ relation terms the mass and charge would make no sense, and thus bare ® particles would exist. But this kind of bare existence is only a fantasy or a ^ strawman easily set and attacked by Bishop Berkeley. More pertinent g examples are perhaps the ghost particles in non-abelian gauge theories, JJ which nobody would take them as real particles although their assumption are crucial for the sake of theoretical consistency (Cao, 1997). I—'
It is clear that scientific realists have to develop an adequate theory of reference to address the Kuhnian challenge. That is, to give scientific revolutions an adequate account without compromising its realist position. More specifically, the theory should be able to lay down the foundation for establishing the reality of fundamental ontology in successful mature theoretical science, rather than for general discussion of what is meant by having a reference, or about how to fix the referent of a name or a term, as discussed and debated by many main-stream philosophers in causal theory, information theory, function or role theory, and description theory of reference.
II. A constructivist account of reference
Among existing theories of reference, causal theory (Putnam, 1975; Kripke,1980) is not adequate for our purpose because of the difficulties with initial dubbing for fundamental ontology, which is usually a hypothetical and unobservable. This is more serious than what Michael Devitt (1981) worried about "the qua-problem"2. Without description, no initial dubbing of the referent for a hypothetical and unobservable entity would be possible. Information theory, including Garath Evens's notion of causal source of information 1982), may not be relevant for the concern of the status of fundamental ontology although the latter is supposed to be the ultimate causal source of information of everything involved in the theory. Function or role theory (Millikan, 1984) is important to establish the initial connection between signs and the signified. Perhaps the ultimate justification about the representational content of any sign, fundamental ontology included, about the world at large, can only be found in the function they served and role they played in the successful dealing with the environment. But the connection may be iconic, indexical or symbolic, not ■g tight enough for a realistic account of the referent of fundamental ontology S) in a mature theoretical science. Theoretical science is too complicated a y representational structure with too indirect connection with the world so ^ that the connection with the world is not direct, and can only be accounted for by properly appreciating various constraints in our conception of reality.
O
>i
8) The major concern of the needed theory of reference is to lay down the
— foundation for arguing that fundamental ontology has referent in the world as it is specified in the theory, rather than the mere claim about functional
5 correlations between the fundamental ontology and reality, which any
jK -
^ 2 By "the qua-problem" Michael Devitt refers to the fact that reference-fixing cannot be a Iflfl purely causal non-descriptive event: in order to fix the reference of a term, one has to know --what kind of object is involved.
e E
empiricists, pragmatists and instrumentalists can accept. This may suggest that the kind of theory was already available in the classic or modernized description theory of reference (Frege, 1893; Russell, 1912; Searle, 1958, 1983), but in fact it is not.
The kind of constructivist account of reference I am going to outline here distinguishes itself from the description theory by two moves. First, it will explicitly address the under-determination problem (Quine, 1960) by appealing to Wimsatt's idea of generative entrenchment (1986). The under-determination problem was not adequately addressed in the description theory, which was thus vulnerable to Putnam's attack in his twin-earth argument (1975). Second, the holistic nature of the description in the general theoretical context of structural realism will be examined and explored by appealing to the intuition of modularity and to the idea of deeper reality as the source of commonality in the unified description of various phenomena. (Cao, 2010)
Let us look at the under-determination first, which may also be referred to as the indeterminacy or openness of reference. In the philosophy of mind, it takes the form of multiple realizablility. There are deep reasons for the inescapable under-determination of theoretical term by description. If we accept that the world is organized as a causal and functionary hierarchy, which is quite reasonable an assumption, then we will face the question of empirical equivalence problem, either causal equivalence, or functional equivalence. A description of a theoretical term, as an observable phenomenon, may be causally related with several causally equivalent agents or entities, or with several functionally equivalent agents or entities. If this is the case, which in fact happens quite often, then from the description, no matter how detail and how exhaust, there is no ultimate justification in principle for uniquely fixing what would be the only true causal or functional agent responsible for the described phenomenon involving the theoretical term, or as the referent of that term. The Ramsey sentence version of ip structural realism cannot escape from this problem. Whatever satisfies C the sentence conditions should be considered as a referent for the ^ theoretical entities specified by the Ramseialized theory. But to satisfy U is to meet the specified relations between the supposed theoretical entity ^ (or entities) and the observations, which does not pose any constraints iq on the internal causal composition or functional organization of the >i theoretical entity. Or as some philosophers, such as William Wimsatt (2007), would say, it puts constraints only on its upward observable ^ relations, but not on its downward compositions for any theoretical g entity sitting at the interface with other theoretical entities and observables which scientists are interested in investigating. A truism ,J® that is often forgotten is that the nature of an entity is always much richer than any description of it. The reason is simple. Many of its '—
Bt O
TIAN YU CAO
properties and relations may not be known to the describers or scientists.
If uniqueness is the defining feature of the reality of theoretical entity then no theoretical entity only satisfying the description could be regarded as real. But it is absurd to take uniqueness of an entity only with its regard to the description. More constraints than descriptive adequacy are required for the reality of a theoretical entity as we will see shortly. When different theoretical entities satisfying the description in certain contexts are compatible with each other, they may sooner or later to converge into one. But they may also reveal themselves as satisfying different descriptions in different contexts. In the sub-hadronic physics of the later 1960s, quarks and gluons are both satisfying the same description of scaling characterizing the behavior of the constituents of hadrons, and thus were named together as partons without any differentiation. But later, differences between charged partons or quarks and neutral ones or gluons were found and different descriptions are needed to label them. (Cao, 2010)
What if the theoretical entities that satisfy the same description but are conflict rather than compatible with each other in their nature? These philosophically interesting cases are frequently cited by anti-realist philosophers. The most a realist could argue seems to say that these entities must be the carriers of some structural features relevant to the phenomena described. This is what some structural realists do. Some philosophers, such as Poincare (1905) and Worrall (1989), the so-called epistemic structural realists, argue that entities can never be known although some structural relations are accessible to scientists. Others, such as French and Ladyman (2003), the so-called ontic structural realists, argue that the structure is the only reality, entities are just metaphors. But then they are not realist of entity. Both kinds of structural realists mentioned above are evasive with regard to a head-on confrontation with the Kuhnian challenge to the reality of fundamental В ontology of scientific theory.
^ Since radical under-determination of the referent for a theoretical
U entity by description without any empirical consequence, as some ^ examples in terms of particle's individuality discussed by Stephen IB French (1998) and his followers, is scientifically uninteresting and can be >i fixed by revising metaphysical scheme, while radical ® under-determination with conflicting empirical consequences can be ^ resolved by further investigations in more differentiating contexts, the only philosophically interesting cases of under-determination are those with compatible entities. .¡s Here I find the idea of generative entrenchment is highly helpful.
pJV According to Wimsatt (1986), the generative entrenchment of an entity in a —' complex system is a measure of how much of the generated structure or
О
activity of a complex system depends upon the presence or activities of that entity. Entities with higher degree of generative entrenchment are more conservative in evolutionary changes of such system. Thus the generative entrenchment acts as a powerful and constructive development-constraint on the course of evolutionary process.
Now science is clearly an evolving and highly complex system. It started with observation, developed into stages with various degrees of unobsevability. Unobservable theoretical entities are introduced on the basis of analogy, attributing identifying features to some hypothetical entities on the basis of knowledge scientists had of other observable or known system. The scientific development in a deep sense is a process of metaphorical extension, with each metaphor taking on more and more reality by its role or function in dealing with the world. (Hesse, 1963) Here I find Ruth Millikan's theory (1984) is highly relevant and useful although I have not found a way to integrate it into my account of reference. But Wimsatt's notion of generative entrenchment can be easily adopted to address the under-determination problem. Since the fundamental ontology in a theoretical science is the one with highest degree of generative entrenchment (all phenomena described by the theory depend on its presence and behavior), it would be virtually impossible to replace it with anything else without changing the whole theoretical description and structure. Uniqueness cannot be ultimately established, but practical uniqueness can be assumed by taking ever increasing number of structural descriptions as identifying features for fixing, or more properly constructing, the identity of the entity. The uniqueness and reality of the theoretical entity can be established, or constructed in a positive sense, this way, to the extent reached by the structural knowledge involving this entity. If the idea of generative entrenchment can be deployed as a strong constraint in arguing against multiple realizability in philosophy of mind, and for the idea that mind can only be a brain phenomenon, then it would be much easier to argue, with the deployment of the same constraint, that QCD as a complicated conceptual scheme can only be realized in quarks and gluons. That is, the reality of quarks and gluons are almost uniquely fixed by the structural U description. ^
Now let me turn to the holistic nature of theoretical description in the iq structural realist context. Term by term reference was rejected by John >i Worrall recently at a conference on structural realism in China (Worrall, 2009), because, he argued, nothing appeared in a scientific theory can be ^ isolated from its global theoretical context or theoretical structure, g everything is only the place-holder in the structure, and thus cannot have any independent existence of its own. Everything is theory-laden, and ,J® thus there is no way to have any independent comparison between a theoretical term with reality out there. The only sense the idea of '—
a
O
Bt O
TIAN YU CAO
reference can have is to talk about global correspondence between a theory as a whole and the phenomena we have observed (the empirical reality).
This position is just an application of Quine's holism (1951). But it is in direct opposition to the reasonable intuition about the modularity of the world. Modularity does not mean that anything can really be detached from its context or not constituted by the context. But the connections within a structure vary in their strengths; some connection are so strong that no separation would be imaginable, others may be quite weak so that there could be quasi-independent existence of some parts in a structure, having only tenuous connection with other parts of the structure and receiving only tenuous influence from other parts of the structure. The most radical holist position would reject any talking about parts. But this is counter intuitive. In my account, holism is taken into account by assuming that the identity of a theoretical entity, fundamental ontology in particular, is constituted by all the structural connection the entity has, which would embrace all the theoretical content, since the fundamental entity is supposed to be the ultimate cause of all phenomena under consideration. Surely this holism with certain modularity being accepted is different from radical version of holism. In the latter version, no parts can be mentioned, no structure can be described in terms of parts, and people would be stuck with the given structure, without any possibility of further investigating the structure in terms of its parts. One may get moved to the parts, but only by taking an instrumentalist stance, viewing parts as a tool for thinking without any reality.
Aside from this negative aspect of the structuralist-holistic account of description, there is also a positive aspect of this account. A structuralist description is a description in which the difference in its place-holders is q"1 discarded and the structural equivalence among them is highlighted. This is jij a very effective way of suggesting a deeper reality underlying the equivalence or commonality. The symmetry argument and the related q ideas about unity and invariance in high energy physics directly led to the ^ suggestion and discovery of sub-hadronic entities is a case in point. (Cao, "g 2010) (B
B)
© III. Implications for history of science | and for descriptive metaphysics
5
.¡A Structural realism as I conceived from the mid-1980s (Cao, 1985,
1997,2003) was a step for a constructivist move in the pursuit of the idea of —' ontological synthesis, which can be adopted to directly address the
Kuhnian challenge to the realist account of science. In my account, structural features described in structural statements were taken to be the constitutive factors or identifying features for fixing the fundamental ontology in theory and, by a realist impulse, also for its referent in the world.
Since the fundamental ontology ofa theoretical science was conceived as being constituted by its structural features described in the structural statements of the theory, since the theory would have undergone changes with the development of science, an unavoidable conclusion would be the replacement of the Aristotelian notion of natural kind by a new notion of "constructed natural kind". While the former in the traditional realism is considered to be pre-given and fixed, only to be discovered and known by us, the latter is conceived as fallible and evolving with the development of science. The constructed nature of fundamental ontology may have weakened the sense of reality, but in fact it does not. All the constructions have to be approved by nature. A case in point is the Higgs particle, which is constructed by theorists. Whether it will be accepted as a fundamental ontology in the standard model and thus as part of reality, depends on whether nature would approve the construction by ways of its responses to what the construction would suggest. So far, no final word from nature has been heard although some claimed that the approval rate has reached 99.999996%. (New York Times, 2012)
Since the fundamental ontology in theory, and thus its referent in the world, is structurally constructed, it would change. Thus no fixity is implied. Partons were the first construction of the sub-hadronic constituents, constrained by the prediction and observation of scaling in deep inelastic scatterings. But soon they were reconstructed as quarks and gluons. If quarks and gluons are real, can we still claim that the partons are also real in the sense that they have referent in the world? Yes, we can and we have to. In some context, partons are real entities. In different contexts, they appear to be quarks and gluons. The constructive nature of the fundamental ontology in theory entails the fallibility of the construction, but, if approved by nature, it also reveals the flexibility in the way the reality is conceived. Does reality has its own way of hierarchical U organization with fixed layers and ontology in each layer? No. Reality is ^ infinitively rich and can be carved in different ways according to human iq capability and interests, subjecting to the constraints posed by nature of >i course. Thus flexibility in theory construction reveals the flexible ways the world can be conceived and described, which, however, has to be approved ^ by nature. No fixity of natural kind can be accepted, all kinds are only g relative to the descriptive scheme, which are real, objective but perspectival. ,J2
This structural and constructive way of looking at fundamental ontology in science has provided a foundation for properly understanding '—
a e
n e
the history of science. Fundamental science in its evolution frequently reshuffles and re-organizes the constitutive factors of its fundamental ontology. I have termed this strategic move as ontological synthesis and first investigated and vindicated it in my historiographical work on the 20th century field theories. I have characterized the history with a synthesis of geometrical structure and quantum entities into a gauge structure. The notion of ontological synthesis gives me a way to address the Kuhnian challenge. Revolutions are accepted since the fundamental ontology before and after a revolution are constructed in different ways. But continuity remains visible and real because the constitutive factors, the structural description of the world, are used in all these constructions, although in different ways.
All what was said above can be summarized as follows: what the constructivist view of reference suggests is that while the historical and constructive nature of fundamental ontology in theory construction is highlighted, the objectivity of the historical construction of fundamental ontology in theory entails that reality is described by our historically constructed, and thus revisable, metaphysical scheme.
References
Cao, 1985 - Cao T.Y.The Intellectual History of 20ül Century Field Theories: an unpublished fellowship Dissertation submitted to Trinity College, University of Cambridge, 1985.
Cao, 1997 - Cao T.Y. Conceptual Developments of 20*11 Century Field Theories. Cambridge University Press, 1997.
Cao, 2003 - Cao T.Y.Can We Dissolve Physical Entities into Mathematical Structures? // Synthese. 2003. № 136 (1). P. 57-71.
Cao, 2010 - Cao T.Y.From Current Algebra to Quantum Chromodynamics -q"1 A Case for Structural Realism. Cambridge University Press, 2010. P. 202-241.
Cao, 1993 - Cao T.Y., Schweber S.S. The Conceptual Foundations and £ Philosophical Aspects of Renormalization Theory // Synthese. 1993. № 97:1. a P. 33-108.
Devitt, 1981 - Devitt M.Designation. N.Y.: Columbia University Press, 1981. Evans, 1982 - Evans G. The Varieties of Reference. Oxford : Oxford University Press, 1982. (B Frege, 1893 - Frege G. On Sense and Reference // Translations from the
^ Philosophical Writings of Gottlob Frege ; P. Geach, M. Black (eds.). Oxford : Blackwell, 1952.
French, 1998 - French S. On the Withering Away of Physical Objects // Interpreting Bodies ; E. Castellani (ed.). Princeton University Press, 1998. P. 93-113.
French, 2003 - FrenchS., LadymanJ. REmodelling Structural Realism: Quantum Physics and the Metaphysics of structure // Synthese. 2003. № 136 (1). P. 31-56.
Hesse, 1963 - Hesse M.B. Models and Analogies in Science. L. : Sheed and Wardm, 1963.
e u
a e
e
E
f
■ H &
23
Kripke, 1980- KripkeS. Naming and Necessity. Cambridge : Harvard University Press, 1980.
Millikan, 1984 - Millikan R. Language, Thought, and Other Biological Categories. MIT, 1984.
New York Times, 2012 - http://www.nytimes.com/2012/07/05/science/cern-physicists-may-have-discovered-higgs-boson- particle.html?pagewanted=all
Poincare, 1952 - Poincare H. Science and Hypothesis. Dover Publications, Inc., 1952.
Putnam, 1975 - Putnam H. Mind, Language, and Reality. Cambridge University Press, 1975.
Quine, 1951 - Quine W.V.O. Two Dogmas of Empiricism // The Philosophical Review. 1951. № 60. P. 20-43.
Quine, 1960 - Quine W.V.O. Word and Object. MIT Press, 1960.
Russell, 1912 - Russell B. The Problems of Philosophy. Oxford University Press, 1959. P. 46-59.
Searle, 1958 - Searle J.R. Proper Names // Mind. 1958. № 67. P. 166-173.
Searle, 1983 - Searle J.R. Intentionality: An Essay in the Philosophy of Mind. Cambridge University Press, 1983.
Wimsatt, 1986 - Wimsatt C.W. Developmental Constraints, Generative Entrenchment, and the Innate-Acquired Distinction // Integrating Scientific Disciplines. Dordrecht: Martinus Nijhoff, 1986. P. 185-208.
Wimsatt, 2007 - Wimsatt C.W.Re-Engineering Philosophy for Limited Beings Piecewise Approximation to Reality. Appendix B. Harvard University Press, 2007.
Worrall, 1989 - Worrall / Structural Realism: the Best of Both Worlds // Dialectica. 1989. № 43/1-2. P. 99-124.
Worrall, 2009 - Worrall J.Defending Structural Realism: or The 'Newman Objection, What Objection? : a paper presented at the Wuhan Symposium on The Philosophy of Science. Structural Realism and the Philosophy of Quantum physics. July 18-20, 2009. Wuhan, China.
a e u
(B >1 a e
o
E
f
■ H &
23
