Varieties of Realism and Anti-realism
Realist Positions:                                                   Anti-Realist Positions:

 Realism with respect to Theories                          Empiricist Anti-Realisms
Entity Realism:                                                       Instrumentalism
Causal entity realism                                              Constructive Empiricism
Evolutionary Naturalistic Epistemology                 Pragmatic Reticulational Anti-realism
                                                                                  Social Constructivist Anti Realism

I. On the side of Anti-Realism:

A: Social Constructivist Anti Realism:

Those who take the rationality crisis as having been settled in favor of a strong version of the "sociological thesis" hold that the task of traditional philosophy of science has been converted into a task for "social studies of science". These non-philosophers hold in effect that scientific rationality is so totally underdetermined by evidence and reasoning (the traditional "scientific method") that only "social factors" can explain actual historical episodes of theory choice. This position rules out "truth" or "reality" as determinants of theory choice, hence it is a form of anti-realism. It may also be described as a form of "relativism" in that it holds that scientific belief is relative to some set of social factors such as economic class, race, gender, etc. It is also referred to as "social constructivism" or "conventionalism" in that it holds that one's world-view is not a function of how things really are, but rather of social factors or "conventions."

Within the community of those who defend the continued role for the philosophy of science, a strong form of the sociological thesis would be unacceptable, but contemporary philosophers may acknowledge some "weaker form" of the sociological thesis by holding that social factors do indeed play a role in constraining the options for theory choice. Such factors will be relevant to constraining which hypotheses (of the infinite number of logically possible hypotheses) are, so to speak, "on the table" for the practicing scientist. However, among those options, the defender of scientific rationality will hold that choice is made on the basis of empirical evidence and reasoning. In particular it is open to some contemporary philosophers to surrender much to conventionalism and yet remain true to the rationally persuasive power of empirical evidence by defending an "empiricist anti-realism".

B. Empiricist Anti-Realisms:

In particular, the empiricist anti-realists party of contemporary philosophers of science, who do battle with scientific realists, can agree with much of the social constructivist critique of (what they regard as) the realists' (naive) view that scientific rationality is a function of the truth about how things really are. However, against the social constructivists, they can stand in favor of a rationality based on empirical evidence and pragmatic values by forging a powerful alliance of the empiricism of the older positivist image and the pragmatists' point that explanation is purpose relative: we accept what works at solving practical problems. These empiricist anti-realists can accomplish this feat by sharply distinguishing between two kinds of statements in the body of rationally warranted scientific beliefs. This is the much discussed observational/theoretical distinction.

Observation statements are, according to this view, indeed true statements about experienced phenomena. This form of scientific anti-realism does not question the mind independent reality of the natural world as did the anti-realism associated with nineteenth century metaphysical idealism. Contemporary philosophers of science are not out to "prove the existence of an external world"; they assume it at the outset. In accepting the mind independent reality of the phenomenal world, these "empiricists" are in fact accepting an ontology, that which is generally called "phenomenalism" or "physicalism"; however, as empiricists, they are uncomfortable when confronted with metaphysical statements about reality. They may be regarded as ontological minimalists who are inclined to accept as "real" only the bare minimum of what is necessary to the rationality of scientific theory choice. Since ultimately all empirical evidence is expressed in terms of observation statements, only these observation statements need be thought of as "true" in the sense of "corresponding to reality"; so only observable phenomena need be regarded as real.

The other body of statements which appear in scientific knowledge, "theoretical statements" are, on this anti-realist reading of theory acceptance, not regarded as "true" in any correspondence sense, so the theoretical terms which appear in such statements need not be thought of as "referring" to real entities, states, or processes behind the screen of observable phenomena. These empiricists accept the logical point that theoretical statements are underdetermined by empirical evidence, but they supplement the gap, so to speak, by pragmatic considerations. They point out that real theory choice is restricted to the field of actual competing rivals, and among these when confronted by equally empirically adequate theories, science chooses those which are most conducive for the pragmatic purposes motivating our theorizing and its empirical testing.

This empiricist anti-realism confronts a serious challenge when it is pointed out that the observational/theoretical distinction is problematic, to say the least. In the first place, the theory-ladeness objection to classical (i.e. old time "empiricist consensus") empiricism seems devastating since it suggests that no statement is purely observational; all are expressed in the language of some conceptual scheme and thus assume some prior "theory" or "world-view." In the second place it appears difficult to define what is "observable" versus what is "unobservable". Obviously if we intend what is in fact spoken of by scientists as "observable" a great deal would be included which is not "observable" in the sense of "experienced" by human perceivers. As empiricists these anti-realists must try to cling closely to a phenomenal sense of "observable," but as pragmatists they are drawn in the opposite direction towards admitting the "observability" of what by the conventions of scientific language are spoken of as "observable" which would not count as "observable" in the narrower empiricist/phenomenalist sense. Finally, it appears that if one allows the technological increase in "observability" to supplement the "naked eye," then clearly "observable/unobservable" is a sliding partition that changes with history. For the realist, these problems all suggest that the O/T distinction will not bear the weight that empiricistic anti-realists want to place on it; however, realists argue that these problems can be adequately addressed by a realist defense of rationality in terms of truth and reality.

Four empiricist anti-realist views:

a) Instrumentalism The oldest form of Anti-realism with respect to both theoreis and entities is instrumentalism.  The instrumentalist holds that theories are tools or rules for calculating numbers of observbable quantities (i.e. they are "instruments" for getting the numbers right).  So construed, theoretical statements are not either true or false, and hence not genuine statements at all.  At least in principle, the instrumentalist is committed to the "eliminability" of theoretical terms:  science could (in principle) be rewritten without any reference to what is non-observable; the fact that we use such theoretical "constructions" is merely for pragmatic purposes.  Every theoretical term is justified in its use only because of the observational consequences which it implies. Thus classical empiricists held that by the appropriate "correspondence rules," (also called "bridge principles") it would be possible to translate all statements employing theoretical terms into statements referring only to observational phenomena.  This view that one could, at least in principle,  "reduce" the theoretical language to a purely observational one was known as "reductionism."  Such a reductionism holds that theoretical terms in our theoretical language only appear to refer to what is unobservable;  they should not be read so "literally." When properly understood they will be seen to refer merely to observational outcomes.
    Anti-realism of this instrumentalist reductionist sort has been largely abandoned because of the futility of attempts to "reduce" (i.e. "eliminate") theoretical language to a purely observational language.

b) Constructive Empiricism (Bas van Frassen):  Van Frassen keeps the anti-realist part of instrumentalism, but discards the reductionist part.  Contrary to reductionism, he holds that theoretical statements really do refer to an unobservable world and are therefore really true or false; they are genuine statements and are properly interpreted literally.  However, their truth or falsity is irrelevant to theory acceptance which is explained by the one abiding aim of all science: empirical adequacy i.e. making correct predictions of observational outcomes (aka: "getting the humbers right").  In agreement with most post-Kuhnian philosophers of science, the constructive empiricist admits that the observational evidence massively underdetermines theory choice, but the actual choices made by scientists are made for pragmatic empirical reasons, not because of all possible theories one of them happens to be true (or closest to the truth). Van Frassen exhibits the fact that "explanation" is a pragmatic notion by arguing that what is considered an acceptable explanation of a phenomenon is relative to the context in which the request for an explanation is posed, and that such a context is itself a function of pragmatic human interests.  Insofar as theoreies are employed tools in constructing explanations, their choice is thus relative to pragmatic interest rather than truth.

    Unfortunately van Frassen's position still needs to insist on those statements whose truth is determinable (by us) and those whose truth is not (theoreitcal statements); thus he is just as much committed to the observational theoretical distinction as sharply drawn as is the instrumentalist.  This seems to be a poor move because a) the o/t distinction is constantly changing in science, and b) it appears to mark a very "anthropocentric" conception of what is observable.

c) Pragmatic Reticulational Anti-realism (Larry Laudan): Laudan criticizes "realism" on the grounds that truth is "epistemically utopian" in the sense that though we may arrive at the truth there's no way to know we are there.  Thus he rejects theory realism, and he seems to infer that this implies rejecting entity realism as well, (though Hacking, for one, would not agree).  Laudan, however, bases his case on the historical record, which, he claims, shows that all aims of science are "negotiable" and in fact do change over time; in this respect he disagrees with van Frassen who holds that empirical adequacy is the essential, universal unchanging goal of science.  Furthermore, the claims that Laudan makes about the goals of science are historically contingent; since there's no way to bar the possibility that in the future perhaps some methods would be discovered by which we could tell whether our theories corresponded to reality, there remains the logical possibility that truth as a goal could become one we could rationally pursue at some future time. Laudan sees all aspects of science as "Heraclitean"; up for possible change, and so is unimpressed by the claim that progress represents a convergence of scientific belief on the way things really are. The "success" of science lies solely in its telling of its own history.

d) The Natural Ontological Attitude (Arthur Fine): Fine develops what he calls a "non-realism" which puts the status of theoretical entities and observational entities all on the same level, but tries to block the further inference that all of these entities, be they theoretical or observational "correspond" to how things actually are.  Fine does not regard his position as "anti-realism" because he admits that one is as rational to believe in the existence of the microentities of theoretical physics (assuming such theories are well confirmed by the evidence), as one is to believe in the ordinary size objects which are the "things" of my everyday existence.  He refers to this commitment as "the natural ontological attitude," abbreviated NOA, and pronounced "knower." Thus our acceptance of the reality of "neutrinos" is no different -except with respect to the empirical means by which we confirm such a belief- than is our belief in tables and chairs.  But this view is not realism either because, Fine holds, one is not justified in further infering from the fact that we accept such "theories," that these theories "correspond to" or "represent" or "mirror" the World, Reality.  Fine clearly rejects theory realism insofar as it depends on such a correspondence notion of "truth," and he does so because he rejects the realist notion that theories are intended to explainanything at all.  Therefore the realist's defense of claiming that realism, as a philosophical theory, explains the success of scientific theoreis by the (alleged) fact that such theories are (at least approximately) "true" is question begging for Fine, because he rejects the view that any theories "explain" anything at all.

II. On the Realist side:

A. Realism with respect to Theories:

Remember that the realist does not deny the anti-realist assertions about theory acceptance as based on empirical adequacy, but claims in addition that the remarkable success which theories of mature science enjoy is best explained by holding that real things at least approximately correspond to the putatively referring terms of such theories. Thus for the realist the O/T distinction does not mark the crucial distinction that it is for the anti-realist: all statements forming accepted scientific beliefs are to be understood as potentially true, and all putatively referring terms are best understood as in fact referring to something, whether that referent is "observable" or not. What is "observable" and what is "not observable" is largely a function of our particular human sense organs, a fact which seems far too anthropocentric to make the O/T distinction the partition between what is "real" and what is merely "constructed."  The fact that observation is "theory-laden" poses no problem to the realist because -as part of the main thesis of realism- the realist holds that such theories are at least approximately true. The fact that they are expressed in a language which imposes some conceptual grid on nature indicates that that framework does in fact, at least approximately, correspond to how things really are categorized in reality. Our conceptual scheme, it is said, "cuts nature at its joints." Thus observations "laden" by such theory (which, by the realist assumption, adequately represents the kinds of things there are in reality) are best regarded as true. Moreover, as we learn more about nature (how things really are) we learn more about how to "observe" entities, states, and processes not experienced by the "naked eye"; so we should expect that this O/T partition is flexible and moves throughout history. Given that scientific methodology (in appealing to empirical evidence) is indeed so laden by theory, how could we explain the success of this methodology for "testing" theory by theory-laden observation, unless the theories with which such methods are laden are at least approximately true?

Contemporary scientific realists also find support in the continuous scale of the dimensions of "theoretical entities." It is indeed rather easy to be skeptical about the entities postulated in "microtheorizing" regarding the structure of matter. Thus no one would regard it as "ridiculous" to question the reality of, say, charmed quarks, although the practicing particle physicist may speak of these entities with the realistic assurance that characterizes the ordinary person's talk about "houses" and "automobiles" and "other people." Nevertheless, one can construct a smooth scale of increasing dimensions to elementary particles, atoms, molecules, macromolecules, cells, microorganisms, dinosaurs, the core of the Earth, galaxies, pulsars, and "gravitational lenses." All of these are "unobservable"; do none of these terms refer? If "electrons" are just "theoretical constructs," is the same true of stegosaurs or black holes? It seems that the anti-realist needs to draw a line somewhere in this continuum to separate the fictions from the non-fictions, the constructural from the real, but -if one admits that what is observable to us is largely an accident of our evolution- it is not clear how or where to draw it.

A recurrent problem for theory realism is its attempt to explain progress in terms of degree of approximation to the truth. One can, as suggested by the pragmatist C.S.Peirce, simply define "the truth" as what an (idealized) process of rational inquiry will eventually arrive at given an infinite amount of time. In this case "progress" converges on "the truth" by definition, but all talk of "correspondence to reality" is then out of bounds. Realists may concede "theory acceptance" to pragmatics, but will want to retain some reference to reality as the grounds for the success of the methodology employed to settle questions of theory acceptance. To do so, realists must maintain that as one theory is replaced by a "better" theory, the successor theory is better than its predecessor precisely because it characterizes the same theoretical entities that its predecessor did, but more accurately, in a way which corresponds more faithfully to how things really are.  Physicists who talk about "electrons" today are talking about the same thing J.J.Thompson ("discoverer" of the electron) talked about over a hundred years ago, but today we have theories which describe that same entity much more accurately than was true in Thompson's time. But this presents a problem, for theoretical terms are defined in terms of the postulates, the assumptions and definitions, of the theory; thus in a change of theory, the meanings of theoretical terms change. But if the meanings of crucial theoretical concepts like "electrons" or "atoms" changes with the advance of theory, then how can later theories be regarded as "about" the same theoretical entities as their predecessors. The realist has to hold that "reference" to "unobservable entities" remains fixed in some extratheoretical manner over the course of the change of theoretical meaning implied by a change of theory. One way realists have employed to keep "reference" fixed in this manner is the effort to develop "causal" theories of reference, but this attempt has not met with anything like widespread success.

B. Entity Realism:

In retreat from problems of reference some realists appeal to naturalism for a justification of their realism, though by no means are all naturalists realists (e.g. Laudan is a "naturalist," but a pragmatic anti-realist). Such "entity realists" will tend to speak less about "truth" of theories and more about the need to assume the "reality" of theoretical entities. Theoretical entities appear as agents in "causal stories" that science uses to explain observable phenomena. In classical empiricism it was a long way between experienced "sensory data" of optical input and such theoretical entities as "photons." However, if, in the spirit of naturalism, we allow our best scientific theories to inform our epistemological accounts of rational belief formation, then causal stories about how the impact of a photon on a cell in the retina causes the perceptual response called a sense datum can provide a scientific justification for holding that photons are as real as the sensory phenomena which even the anti-realists do not doubt. Moreover, as we learn more about such entities, i.e. as we replace today's theories by future theories, we will be able to explain this process in ever greater detail thus providing increasing justification for regarding the causal agents appearing in such scientifically informed accounts as real entities. Advances in empirical sciences such as perceptual psychology, cognitive science, and neurophysiology all combine to increase our confidence in unobservable entities as real causes for observable processes.

Two types of entity realism:

a) Causal entity realism (Ian Hacking): The turn away from "theory realism" (theories are true representations of an unseen reality) towards "entity realism" (we have good reason for believing in the reality of theoretical entities which play a role in causal stories about observable phenomena) in philosophy has breathed life into an aspect of science formerly neglected in philosophy of science, and that is the role of experiment. Particularly in the case of microphysics, anti-realism seems most plausible at the level of the theoretical representation of the quantum mechanical states of "fundamental particles." However, when it comes to experiments in which we manipulate these particles by beaming them along various paths and making them collide with parts of "detectors," then a causal entity realism seems most attractive. This line of reasoning, whatever its merits as a defense of realism, has had a salutary effect in philosophy by balancing the former overemphasis on theories with a growing literature displaying a new found enthusiasm for the analysis of the role of experiment in settling questions of theory choice.

b) Evolutionary Naturalistic Epistemology Some naturalist-realists also appeal to the science of evolutionary biology to advance the argument of "evolutionary epistemology" which justifies their claim that the evolution of human perceptual faculties through the process of natural selection has selected in favor of organisms that come hard-wired with belief formation faculties which tend to produce true beliefs over those that tend to produce false beliefs. If out of all logically possible conceptual schemes, humans come "hard-wired" with certain ones, we have grounds for regarding those conceptual schemes as corresponding to how nature is actually divided up (in reality) rather than as arbitrary conventions of our conceptual schemes. Unfortunately this line of argument must confront the problem that such non-epistemic virtues as speed in eluding predators or protective camouflage may have greater survival value than epistemically accurate faculties. The survival success of cockroaches over humans is a sobering caution against overstressing evolutionary arguments for the truth of beliefs reached by processes which have originated through biological evolution.  Skeptics of evolutionary epistemology would point out that human beings have been around for such a relatively short time, evolutionarily speaking, that it is rash to reach any conclusions about the survival value of human consciousness, much less that it has evolved in a way so as to represent reality as it truly is.