The Meaning of Realism
The task of saying what realism is is not a trivial one. We need an account that generalizes across all areas in which philosophers have found it natural to speak of realism and anti-realism. So the account must be abstract and topic-neutral not restricted to one or two specific examples. A standard attempt invokes the notion of mind-independence: to be a realist about a certain subject matter is to hold that this subject matter is mind-independent. The trouble with that formulation is that it doesn’t apply to realism about the mind itself—it is not mind-independent! Also, why exactly is this thought to be the import of realism? How can mere independence from the mind constitute realism? It might be thought that we need to bring in the notion of existence explicitly: to be real is simply to exist. But that doesn’t work either because even an anti-realist believes that the subject matter in question has existence—for example, material objects exist for an anti-realist just because they are constructions from sense experiences, and they exist (similarly for behaviorist anti-realism about the mind). The intended meaning of realism in these cases is surely something like the following: material objects exist apart from sense experiences (as mental states exist apart from behavior). The mind-independence idea was not wrong for realism about material objects but only as a general rubric; instead, we can generalize by saying that realism about the mind takes it to be independent of something else, namely behavior. Schematically: to be a realist about X is to hold that X exists independently of Y, where Y is the thing an anti-realist identifies with X. Thus to be a realist about material objects is to hold that material objects exist independently of sense experiences, while to be a realist about mental states is to hold that mental states exist independently of behavior. The realist holds these things to exist separately, while the anti-realist denies this: separate existence is the crucial point. Common sense affirms such separate existence and philosophical realism endorses it—while anti-realism disputes the thesis of separate existence. Realism thus asserts ontological plurality while anti-realism denies plurality (e.g., material objects are not different in kind from sense experiences). A metaphysical realist is a metaphysical pluralist: she insists that there are two things at issue not just one.
Here we glimpse an essential feature of the realist position: the world is conceived as laid out in separate compartments connected in various ways, where these compartments exist apart from other compartments. That is the pluralism characteristic of realism: objects are not experiences; minds are not bodies. The realist is a pluralist separatist. The same structure is at work in other areas: properties (universals) are not predicates; moral values are not empirical facts; numbers are not numerals; possible worlds are not mental constructions. The global realist thinks that the former categories exist separately from the latter categories, while the global anti-realist asserts unity and assimilation. The realist denies the assimilationist tendencies of the anti-realist—that is the essence of her position. Similarly for realism with respect to time: to be a realist about the past or the future is to hold that the past and future exist separately from the present—that they are distinct compartments of reality. The anti-realist contends that past and future can be viewed as somehow aspects of the present—constructions from it (memories, expectations). Even in cases where there is ontological dependence, the realist will claim that there is ontological separation: the mind may depend upon the brain but it exists separately from the brain, i.e. it has its own mode of existence (its own irreducible properties). Thus a materialist cannot be a realist about the mind—not really. He thinks there is one thing not two, so he denies separation. This is why monism always turns out to be a form of anti-realism. Take a case where anti-realism looks plausible—say, about color or fictional characters. Here the realist would double the ontology: in addition to dispositions to appear and creative mental acts there are alsocolors and fictional characters—they exist separately. This strikes us as too much pluralism: colors are just projections from appearances and fictional characters are just another way to talk about an author’s imagination. It is the thesis of separation that makes the difference between a realist view and an anti-realist view: the idea of ontological division, or the lack of it. If you were to draw up a big map of Reality, you would find different continents located in different places, according to the realist; by contrast, the anti-realist contends that there are fewer landmasses than we tend to suppose—it might all be Greenland, say.
I just used the idea of spatial separation to explain the meaning of realism, and indeed I think this is the root of the general notion. Spatial separation is the paradigm expression of realism: when things exist in separate places they cannot be assimilated and must be regarded as existing independently. Let me illustrate this idea with a model case designed to bring out the conceptual structure of realism and its opposite. Consider a fishpond with surface swimming fish and a chalky bottom. People like to gaze into the pond and follow the antics of the fish, a lively reddish species. But there is something curious about this pond: there are also fishlike shapes moving around at the bottom of the pond, though they are difficult to see clearly. What is odd is that their movements are synchronized with those of the surface fish, as if the two are joined invisibly together: there are correlations, predictable dependencies. There are two schools of thought about these bottom-dwellers: one school holds that we have here a second species of fish that mimics the behavior of the surface species, perhaps for reasons of camouflage; the other school believes that there is only one species in the pond and these fishlike shapes are merely shadows of that species (a third group maintains that they are actually after-images caused by looking at the brightly colored surface fish). That is, one school believes in a separate species of fish in addition to the species open to plain view, while the other holds that the flitting shapes are just shadows cast by the fish they can see (or after-images of them). According to the latter, there is just one species of fish in the pond not two (no fish pluralism) and a kind of natural error occurs as a result of how the light is cast onto the bottom of the pond. According to the former, we have a separately existing species that happens to lie somewhat out of sight, but is no less real for that. The two schools of thought label themselves “realists” and “anti-realists” about the shapes glimpsed in the depths. The realists believe in a duality of separately existing species, while the anti-realists insist that there is only one species of fish in the pond, though our perceptions lead us to postulate two (we mistake shadows for fish). Intuitively, the issue turns on whether the shapes on the bottom have an independent existence relative to the fish near the surface. Let’s suppose that they don’t and the shadow school is right: then anti-realism turns out to be true–fish monism, no separately existing species.
The realist view has it that there is a spatially separated species of fish in the pond: that is what makes their view realist. The anti-realists deny that there is any such species existing in a separate region of space: that is what makes their view anti-realist. And surely it is sufficient for realism that entities of a certain kind exist separately in a certain region of space. But is it necessary? Does the issue always turn on the contents of regions of space? In the case of material objects we can certainly say that realism is the view that material objects exist in space separately from sense experiences (which may or may not be in space); but what about realism concerning the mind—should we say that mental states exist in space separately from behavior? That presupposes that the mind exists in space, and realism about the mind surely does not depend on making that assumption. However, it has been common to formulate the matter in terms of subjective space and objective space: does the mind exist in a subjective space separate from objective space? If so, we have mental realism; if not, we have mental anti-realism. The important point is that the mind is conceived as occupying a quasi-space that exists alongside physical space—and this is what qualifies the position as realist. There is some sort of realm or region in which the mind has its being, and this realm or region is not the same as the realm or region occupied by the body (compare the bottom of the pond and the surface of the pond). Thus the situation is analogous to the situation in the pond: regions of physical space form the paradigm, but we can extend the notion of spatial separation beyond the simplest kind of case. The mind does not occupy the same kind of space as the body, but it exists in something analogous to that space, something that allows the notion of separation to gain purchase. The heart of mental realism is the thesis that the mind exists in a part of reality separate from the part occupied by the body—in a separate space in the simplest formulation. It is not that the mind coincides with the body (as the external world does not coincide with sense experience): that would deliver anti-realism. Realism involves parallel existence not single existence.
But there are more difficult cases to contend with. What about Platonic realism? The allegory of the cave fits the general idea nicely: the world of forms exists outside the cave and can be reached only by an arduous climb—here the idea of spatial separation is rendered explicit. An anti-realist about universals would hold that there is nothing more to so-called forms than the shadows cast on the walls of the cave, no separate realm existing at some remove. In the cave allegory Plato even supposes that we could travel to the region of space in which the forms exist. But this is just a metaphor—how does Plato literally conceive the existence of universals? Well, the idea of Platonic heaven is regularly invoked: not another region of physical space, as in the cave allegory, but an analogous quasi-space. Without some such conception it is hard to see how Platonic realism could be given cogent content: for how could we make sense of a separate realm populated by universals except by means of some quasi-spatial way of thinking? The anti-realist will certainly insist that such a conception is deeply mistaken (while accepting that it captures what the realist has in mind): there is nothing to talk of universals except the earthbound language in which we describe things (the familiar space of linguistic use). So we can accommodate realism and anti-realism about universals by adopting some natural extensions of the basic idea of spatial separation.
What about moral realism? Here things take a murkier turn and even metaphors are in short supply. In what sense are moral values for a realist located in some other region or type of space? Could we travel to the place in which they independently reside? Is there some moral quasi-space that houses them? Maybe we could devise a “parable of the pearls” according to which moral values exist in a splendid museum of shiny baubles, but no such ideas have gained traction in the history of thinking on this topic. Maybe the quasi-space of the divine might be recruited to contain moral values, but the moral realist surely doesn’t want to be committed to anything like that. All we have is the rather thin idea of the fact-value distinction—but nothing to give substance to the idea of separate existence. There is no grand pond at the bottom of which values might languidly swim. The anti-realist is thus tilting at windmills to some extent, finding nothing to get his dismissive teeth into. By the same token the moral realist is left with a rather schematic thesis; all she can say is that values are not reducible to facts. Perhaps this is because we are dealing with metaphysical questions about values, unlike the other questions that attract realist and anti-realist rhetoric. My own imagery in this area tends towards depicting values as some sort of iridescent color unlike any color seen with the human eye—with a softly glowing quality. But this is fanciful stuff.
Realism about modality requires yet further extensions of the spatial paradigm. The idea of logical space presents itself. The actual world exists in one part of logical space, but possible worlds extend out from there across logical space. To be a realist about necessity is to accept that necessity resides in the existence of the space of possible worlds—a space removed from that of the actual world. The modal anti-realist, by contrast, holds that talk of necessity adverts to nothing beyond the actual world—there is no separate logical space in which modality has its being. The enormous size of logical space is testament to the robustly separate reality occupied by necessity and possibility; it is a very large pond in which possible worlds swim. Or maybe we should compare modal realism to what might be called “galaxy realism”: the thesis that those patterns of light we observe in the night sky are not just local optical phenomena in our atmosphere but remote collections of stars and planets just like our sun and its planets. The astronomical anti-realist denies the separate existence in space of other stellar systems, while the realist asserts that space is replete with star systems every bit as vast as ours. Similarly, our talk of necessity and possibility indicates no world beyond the actual world for the anti-realist, while the realist insists that a vast totality of possible worlds extends far out into logical space. Modal reality, like astronomical reality, consists of enormous objects made up of innumerable parts, according to the realist; while the anti-realist insists that there is nothing beyond the local and observable—no other possible worlds but this one, and no galaxies apart from the one we live in. Again, it is the idea of separate existence that captures what is at issue.
We should not expect the issue between realists and anti-realists to be capable of rigorous formulation. These are intuitive and impressionistic concepts deployed to correspond to metaphysical pictures. Still, we should at least try to elucidate what constellation of ideas underlies these labels, aiming to articulate what is driving philosophical thought. The idea of separate existence in space, suitably extended and qualified, seems to be at the heart of the dispute between the realist and the anti-realist. 
 Could this be why moral realism has comparatively few takers? We just don’t have a vivid picture of what it would be for it to be true. The case contrasts sharply in this respect to realism about material objects. (None of this refutes the claim that it is true.)
 The background to this essay is Michael Dummett’s attempt to find a definition of realism adequate to the full range of metaphysical debates in which that notion has been employed. He tried to make bivalence into the touchstone of realism; I think we need a richer and more metaphysically substantive notion if we are to do justice to the intended meaning of the word “realism”.