CHAPTER IX

REPRESENTATION AND REALISM

In this final chapter I will attempt to spell out the epistemological implications of the theory of representation presented in this dissertation and see what type of realism it tends to support. If I were inclined to name the type of realism defended here (one most often names a position in order to attack it), I would use the name Richard Rorty applies to the views of Thomas Nagel, Intuitive Realism.[1] (Rorty 1982) I would take three central theses as characterizing intuitive realism. First, the world that we talk about has a nature independent of our characterization of it. Second, since we know the world only by representing it, our knowledge, by its very nature, is incomplete. Third, although knowledge from particular perspectives is necessarily incomplete, it can be objective and true, so far as it goes. It seems to me that the term Intuitive Realism is appropriate, because its main thesis is that objective knowledge is possible from particular perspectives. It is a rehabilitation of intuition as an essential form of knowledge. By intuition I mean any knowledge or representation from a single perspective. (I believe this is generally what we mean by intuition in both the Kantian and everyday senses.) Not all knowledge is reflective, involving the comparison of different perspectives, and the possibility of any knowledge at all rests upon knowledge from particular perspectives.

I will begin my consideration of the implications of an alternative theory of representation by looking at how the theory can be used in a critique of the assumptions that lead to Putnam's realism.

9.1: Putnam and the Perspectivist Fallacy

We saw in Chapter Eight some of the main assumptions that led to Putnam's internal realism were. Let me review them quickly: It was assumed that representations were objects or signs that gained their reference through our interpretation of them. It was assumed that we only experience representations, that we have no direct access to the world. We also saw that Putnam assumed something like a subject-predicate or substance-attribute metaphysics; that is, he assumed that the properties or characteristics in our representation of objects really do belong to the object, as these properties are represented. He was committed to some version of self-identifying objects, and when faced with the choice of abandoning self-identifying objects and accepting ontological relativity or abandoning externalism, he abandoned externalism. He says: "... this story may retain THE WORLD but at the price of giving up any intelligible notion of how THE WORLD is." (Putnam 1976, p. 132) We saw that this was due to his position on reference. It seems an obvious fact that we do determinately refer, and we could not be referring if we contact objects only mediately, through our representations. Therefore, the things to which we refer must be within our representational systems.

This argument depends for its force upon the strong version of the perspectivist fallacy; it depends upon the thesis that if a representation is from a particular perspective, then it cannot refer to anything outside of that perspective. Putnam states the thesis in this way: "If one cannot say how THE WORLD is theory independently, then all talk of these theories as descriptions of 'the world' is empty." (Putnam 1976, p. 133)

We saw earlier that the perspectivist fallacy is a result of the attempt to make representations come alive. It arises from a misunderstanding of what representations are and how they function. In particular, it is a misunderstanding of representational self-reference. It is essentially a result of taking representations to be objects or signs. Once one sees that such objects cannot guarantee their own correspondence to the world, that they cannot represent by their intrinsic similarity to their objects, then objective knowledge will be seen to a require a step outside of the present representation to other representations from different perspectives. Putnam diagnoses the problem well and traces it back to Occam:

The problem, in a way, is traceable back to Occam. Occam introduced the idea that concepts are (mental) particulars. If concepts are particulars ('signs'), then any concept we may have of the relation between a sign and its object is another sign. But it is unintelligible, from my point of view, how the sort of relation the metaphysical realist envisages as holding between a sign and its object can be singled out either by holding up the sign itself ... or by holding up yet another sign... . (Putnam 1976, pp. 126-127)

The point at which the physical-visual model of representation begins to unravel is representational self-reference, where one begins to be concerned with the relation between the representation itself and the object, and where one begins to be concerned with justification. As Rorty saw so well, the traditional model of representation begins to run into problems with the Platonic distinction between opinion and knowledge, with the insight that knowledge is not just true belief, but true justified belief.[2] With this the problem of representational self-reference becomes central. Knowledge requires not just a representation of the object, but a representation of the relation between the representation itself and the object. It was quickly seen that self-justifying representations were impossible on the traditional model.[3] Objects do not determine their own relation to the world, and when they are taken to be representations they form a barrier between the knower and the world. Therefore, there can be no knowledge from particular perspectives or representations on the traditional view; justification always requires another representation from other perspectives. Since a sign cannot determine its own relation to its object, it cannot represent its own relation to its object; it cannot justify itself. Justification always requires another representation, which itself requires justification.

Putnam's version of the perspectivist fallacy is his argument that reference requires independent access to both the sign and its referent. A sign cannot determine its own referent, so reference requires a representation of both the sign and its object in which the sign can be assigned to the object or interpreted.

We saw, however, that if representing is seen as an activity of interaction with the world then it becomes possible to understand how there can be self-justifying representations. Representations as described in Part Two not only represent an object, by connecting modes of interaction and attributing them to the object as their common causal origin, but they also represent the relation between the act of representing itself and the world, by being an act of connection that can be caused by the interaction with the world. We saw that representation involved the application of concepts, or active tendencies to connect modes of interaction with the world, and that interaction with a domain of interest in an act of representing could reinforce or inhibit the connections made by the concept in a way that was felt as satisfaction or dissatisfaction of the tendency. The connection of the modes of interaction represents the domain as their common causal locus, and the felt satisfaction represents the act of connection itself as being caused by interaction with the domain and not by the subject. Thus, on this view, all representations are inherently self-referential.

This is possible because representations are not seen as objects or static entities that somehow are put in some relation to objects; they are not seen as symbols. Rather, representations are seen as acts involving an interaction between the subject and the object. Thus, they can interpret themselves (or, to put it more precisely, they are acts of interpretation), since they are an act of isolating a domain by connecting modes of interaction with that domain. They refer to themselves, because they are the expression of teleological tendencies that aim at their completion. And they can justify themselves, because they involve causal interaction with the domain to which they refer in which the tendency they express can be satisfied or frustrated. On this view objective knowledge is possible from within a single representation, because the representation is seen as an activity and not an object.

The perspectivist fallacy, then, is mainly a misunderstanding of how representations work due to taking physical objects or symbols as a model for representation. Since such representations cannot come alive and determine their own interpretation, they cannot justify themselves. Therefore, knowledge will always require stepping outside of one's perspective and forming a more general representation. But representations are not objects and they interpret themselves. Since they include the interaction of subject and object, they can contain their own justification. The perspectivist fallacy, then, is due to a failure to see how representations can be (and essentially are) self-referential.

This is essentially the problem with Putnam's argument that determinate reference requires independent access to both sign and object. By taking representations to be objects (even objects that are only interpreted by us), he makes the self-reference that is essential to representation impossible. His internalism comes about as a last ditch effort to save a model of representation and the determination of reference that his own arguments show to be deficient, rather than as an alternative account of representation. (Making objects internal to the representational system allows non-self-referential representations to refer.)

Before we see how this last ditch effort itself fails, let us see some more implications of the view of representation presented here and how it requires a revision of some of the presuppositions that led to the dilemma faced by the traditional model.

9.2: Representation, Truth, and Generality

One of the most important implications of the view presented here is that our knowledge is essentially incomplete. The view here is that all knowing is through representing and that representation from particular perspectives can be objective and forms the basis of all our knowledge. This view also holds that the content of all representation lies in the connections made within it, not in the properties that are connected.

This view implies that our knowledge is limited in two ways: First, we never know objects as they are in themselves. The properties we perceive are the felt modifications that the object produces in us; it makes no sense to attribute them to objects themselves. We can only know objects as the causal locus of some set of properties. We can only know that certain domains have causal powers which are manifested as the properties we feel. (This limitation is the topic of the next section.) Second, (and most important for this section) a representation from a particular perspective, though true and objective, is tied to that perspective and cannot necessarily be generalized to other perspectives. All representations are particular, they connect properties as stemming from a particular interaction with a domain. All generality is a matter of concepts, or tendencies to represent certain domains in certain ways. But it is only representations that are self-justifying; only they can correspond to the world. There is no way of being sure that our concepts will always produce satisfaction whenever they are applied to other domains.

One of the main sources of the traditional model of representation and the troubles that go along with it is the tendency to see this limitation as a problem with representation. One of the sources of epistemological dilemmas is the confusion of generality with truth. Only representations are true; only they co-respond with the domain with which they are interacting. But it turns out that truth (by itself) is not very important, if what you want is to control and predict the world. Just because two properties were connected in one domain, does not mean they will be so connected in other domains. What is important for the purpose of controlling the world is the ability to apply the connections one has made in the past to present situations in the correct way. But this is not a matter of the correspondence of any particular representation with the domain with which it interacts. Just because my representation of the cat as being on the mat is true now, does not mean it will be true five minutes from now. Generalizability is not a matter of truth of representation. The truth of a representation guarantees only that the connection made is caused by the interaction with the present domain; it does not distinguish between those connections which are contingent and those that are necessary.

Generalizability requires the differentiation of essential connections from contingent ones. In order to know which foods are bad to eat or poisonous, one needs to know more than just whether a particular instance of the type of food was bad to eat one time; one needs to know if the bad symptoms are essentially connected to that type of food. This not a matter of having a true representation. It is a matter of having concepts which tend to produce true representations.

Remember that concepts involve tendencies to connect certain properties and they also involve sets of tendencies to orient our bodies so as to directly refer to certain domains. Let us call the first type of tendency representing tendencies and the second type referring tendencies. Generalizability, then, rests upon the ability to form concepts in which the representing tendencies are connected with referring tendencies that lead to interaction with domains in which the connections made by the representing tendencies are satisfied.

If we imagine that concepts tend to reinforce those tendencies that produce satisfaction, then concepts, in their normal operation, will tend to become generalizable. The possibility of this will rest on our ability to have true representations, but the generalizability of concepts is not truth.

Generalizability is something we arrive at only due to pragmatic considerations. It is a result of our tendency to reinforce connections that produce satisfaction. Our knowledge is inherently limited in this respect. The world only determinately constrains us in particular acts of representing. Our concepts are constrained by the world only insofar as they lead to these acts. Our knowledge, therefore, is limited to those perspectives and domains that we can isolate out given our modes of interaction with the world. But this is not due to a failure of our representations. It is not a result of a veil of ideas that stands between us and the world. It is a result of the fact that representations are interactions with the world. Since representations are embodied on this view, the generalizability of their application in concepts will be constrained by the way they are embodied. Since representations are ways of interacting with the world, our knowledge will be limited by the ways in which we can contact the world through our bodies and sense organs.

The attempt to transcend these limitations leads to the problems of a perspectivist model of objectivity. This is the confusion of generalizability with truth. The attempt to see representation as something that goes on by itself, apart from our embodiment and interaction with the world, makes it necessary to see truth and objectivity as things that must be independent of particular perspectives and ways of interacting with the world. This attempt to identify generality with truth is self-defeating. For particular perspectives and acts of representing cannot be general. Therefore, on this view, they cannot be objective or true. The only source of objectivity, however, is particular interactions with the world (only in these interactions are the connections made by our concepts constrained by the world); but since these are merely subjective on this view, we are left without any source of objectivity but internal constraints. The incoherence of the image I used for the perspectivist model is instructive here. In attempting to so increase the number of posts holding up a platform that the amount of weight supported by any one becomes zero, you are taking away the foundations of the platform rather than strengthening it. If, in the end, only completely general representations can be true or objective, then the particular perspectives upon which these general truths rest are worthless as foundations.

Retaining the distinction between truth and generality (that is, the distinction between representations and concepts) is essential. Only then is it possible to see how we can have general knowledge (concepts that are generalizable) that rest upon objective foundations (true particular representations). The price of this, however, is the admission that our knowledge is necessarily limited by the modes of representing allowed by our type of embodiment.

9.3: Properties and Objects

We have seen that Putnam is committed to a substance-attribute ontology. We now need to see what type of ontology is required by the view presented here. This will involve seeing what view of properties and objects these views lead to.

We have seen that on this view the content of representations lies in the connections made within them and not in the properties which are connected. Properties are simply the felt character of an interaction with a particular domain through a particular mode of interaction. There are two reasons, which have become commonplace since Berkeley and Hume, why it does not make sense to attribute properties to objects themselves and not just to our interactions with them: First, the same properties can be felt in the absence of interaction with the object (through imagination and memory). We saw that this was due to our ability, through our concepts, to stimulate modes of interaction independently of their activation from peripheral neurons. Second, properties are tied to particular modes of interaction and the particular sensory and bodily connections that define those modes of interaction. Different organisms from different perspectives with different sensory apparatus will perceive different properties. For these reasons, it does not make sense to attribute properties, or the way an object feels in a particular mode of interaction with it, to objects. Nor, therefore, does it make sense to take the properties connected in a representation as part of its content.

For example, the property redness is the felt character of interactions with certain types of objects through our eyes and the neural apparatus connected with them. The same objects manifest different properties when interacted with according to different modes, with different sensory organs. There is no reason to consider these properties as parts or attributes of the object.

We sometimes talk as if we are attributing properties to objects, or as if we take properties as expressing objective characteristics of objects. For example, we say that roses are red or that redness is a particular wavelength of light. Insofar as these statements have objective content, however, it lies in the connections made between redness and other properties, such as the other ones that allow you to pick out roses or certain wavelengths of electromagnetic radiation. In making these statements, we are not attributing redness to noumenal roses nor identifying it with a property that the object has independently of any mode of interaction. Rather we are connecting redness up with properties manifested through other modes of interaction. (This explains why primary properties, which are manifested by more than one sensory mode, were sometimes taken to be objective. They are not single properties; they are ways of representing objects as being the common causal locus of two sets of properties, usually visual and tactile.)

A substance-attribute or subject-predicate metaphysics is tied up with the physical-visual model of representation. If representations must be similar to their objects in order to represent, then the properties that make up representations must have their counterparts in the objects. Furthermore, if representations generally have a subject-predicate structure, then objects must have a similar structure in order for our representations to mirror them.

The subject-predicate structure of some representations, however, does not reflect the necessity of mirroring a world with a substance-attribute ontology. Rather it results from the fact that the reference of our representations must be determined independently of the correctness of our representations in order to allow for the growth and correctability of our representations. Thus representing often involves connecting one set of dispositions to enter into acts of referring (the subject) with another set of dispositions to represent a domain in a certain way (the predicate). This does not involve attributing the properties involved in the predicate to the domain; it is simply representing the domain as the common cause of the set of properties. The properties themselves are simply the felt effect of the activation of our modes of interaction by the domain. There is no reason to think that the structure of the object mirrors the structure of our sensory apparatus. The substance-attribute metaphysics seems to be due to a misreading of the significance of the structure of representation.

This misreading is one that would be natural for one who held that representation occurs through the intrinsic similarity of the representation to its object. It is also a fortunate misreading, for it allows us to have knowledge of objects as they really are. For it so happens, on this view, that the properties we perceive objects to have through our senses really do inhere in the object, even though the substance in which these properties inheres becomes a mysterious 'something I know not what'. This implication of the view makes it powerfully attractive for those like Plato and Aristotle, or Kant and Putnam, who take the fact that we have knowledge of the world as given or as a starting point. For this model makes possible a knowledge of how the world really is, although it does require a reinterpretation of what the world is in order to allow this. (In Plato and Aristotle, the world is seen as containing universals to become more like our concepts; in Kant and Putnam the world become a construct of our concepts in order to allow determinate reference or objective validity for our concepts.)

Unfortunately, the only reason to hold such an ontology would be if one held that we represent through the similarity of our representations to their objects, a view that would not work even if it were possible. The price of giving up a substance-attribute ontology, however, is giving up our ability to know the world as it is, to mirror the world with our representations. The price of retaining this is to make the world a construct of our concepts, as Putnam sees. He chooses an Oatmeal theory of the noumenal world in order to retain objects with properties and the ability to refer to them. We now need to see what view of objects one can have if one gives up substance-attribute metaphysics.

The view of representation given here requires that the world have structure apart from our representation of it. The world must have sufficient structure to constrain the connections made in acts of representing in order for this view to work. Domains must either be or fail to be the common causal locus of properties connected in acts of representing. To be an object on this view is not to be a substrate with various properties, it is simply to be the common causal ground of a set of properties (on this view, these properties are the felt character of the interaction with a domain according to certain modes). The properties are causal effects on the subject seen as springing from a common source, rather than attributes that inhere in the object and which produce images of themselves on our senses.

On the substance-attribute view, objects are defined independently of the particular properties they manifest. There is a substantial structure that defines the object upon which attributes can be pinned. On the view advanced here, however, objects are defined by the set of properties of which they are the common causal origin. To be an object on this view is simply to be the common causal nexus of the activation of a set of modes of interaction. Discontinuities in the connections between properties experienced in interaction with domains is what defines objects. A domain which is an object causally reinforces the connection of modes of interaction in a way in which the environment surrounding that domain does not.

But this makes the definition of objects dependent upon the modes of interaction operating. A domain may connect one set of modes in a way that is discontinuous with the environment, while the way it connects another different set of modes of interaction may be completely continuous with the environment.

For example, a completely transparent object is continuous with the environment with respect to interaction by sight, while discontinuous with respect to touch. An opaque gas may be just the opposite: continuous with respect to touch and discontinuous with respect to sight. What is recognized as an object (i.e., as a domain connecting properties in a way discontinuous with its environment) will depend on how one is interacting with the domain. We take things like doors and desks to be objects because our bodies are so structured so that our bodily interactions with these connect properties in a way discontinuous with the environment. But a Martian whose body structure allowed it to pass as effortlessly through material objects as we pass through air would not recognize the same objects as us. (At least not by means of these tactile interactions.) Whether a domain is recognized as an object depends on how one interacts with it.

But this is not so because objects are constructed by us in the act of representation. Objects are discovered through representation, not constructed. The very nature of representation, on this view, precludes an Oatmeal theory of reality. Representing is the product of concepts which are tendencies that can be satisfied or frustrated. The satisfaction or frustration of particular representations is unintelligible apart from the constraint of external objects with structure. Concepts are tendencies that aim at their own satisfaction; they will not construct objects that frustrate them. A world that constrains our conceptual activity in this way must have structure. It may not have a substance-attribute structure, but a world in which domains connect modes of interaction in ways that are discontinuous has structure.

If to say that the world has structure is to say that there are discontinuities in its causal powers that from closed regions of space-time, then to say the world has mind-independent structure is to say that there are mind independent objects. Objects are defined by closed discontinuities in the causal powers of the world. The objects that we perceive objectively must be independent of our perception of them.

This, of course, does not mean that we have some non-conceptual access to objects. The exercise of our concepts in acts of representing is our access to objects, and the objects we perceive depends upon what concepts we use. Yet, the objects exist apart from our perception of them; the concepts determine only which of the pre-existing objects will be perceived.

We experience the structure of the world only through its causal effects on us. This is what it means to say that we know only by representing. To say that the world consists of objects is simply to say that there are discontinuities in the way the world connects the various causal effects it has on us. It is not to say that there are determinate substrates onto which properties are grouped. Within a single domain there are many possible objects. Which ones are picked out depends on the modes of interaction with the domain. The properties define the substrate; which is to say that there is no substrate (a substrate that changes with the properties is a contradiction in terms), only the myriad of structures that connect the various modes of interaction. The world, though structured, does not cut itself into exclusive discrete and isolated units. The objects or closed discontinuities of the causal powers of the world overlap in space and time. We isolate certain of these objects or closed discontinuities by representing the world according to only some of the possible modes of interaction with it.

An image that I find helpful in understanding this view is of the runway lights of a large airport at night. There are a number of pairs of parallel rows of lights, each pair defining a runway. These pairs intersect each other at various angles. If one views these lights from ground level on a plane moving into position to take off, they appear only as a tangled clump of lights as long as one is not looking in a direction close to parallel to any of the runways. As one moves to a position where one's line of sight is close to parallel to a runway, the structure of the runway emerges. A pair of parallel rows of lights pops out from the morass of lights. That perspective on the airport allows one to see that runway, while hiding others from view. As one moves around the airport, changing perspectives, different structures appear and disappear from view. These structures are not created by the perspectives or by our activity; they were always there. One simply had to approach them in the right way to discover the structure.

In the same way, the world contains many discontinuities or structures, in which lie the possibility of the discovery of many objects. One needs simply to interact with the world in ways that allow the discovery of these objects. It seems that this is basically what science does: it attempts to find new ways of approaching the world in order to get it to show its structure.

The view presented here is essentially a version of the ontological relativity that Putnam rejected because it retained the world while abandoning any intelligible notion of how the world is. What objects one recognizes depends on the modes of interaction one has with the world. This view certainly abandons any possibility of knowing how the world is apart from its causal effects on us. It is a reluctance to admit that the world may have a structure that our representations cannot mirror that accounts for the resistance to ontological relativity. Yet the knowledge we have of how the world from its causal effects tells us what discontinuities there are in the world and what properties are connected as causal powers of the domains defined by these discontinuities. This is certainly knowledge of how the world is. It is at least enough for efficient control and adaptation to the world.

In this section, I have been explaining what ontological views are presupposed by the views of this dissertation. Views about how the world is apart from our representation of it cannot be defended by arguments concerning the nature of representation. There are two possible defenses of such ontological views. The first type of defense is empirical. Our empirical knowledge certainly seems to support the view that the world has a structure. We have gone a long way towards discovering some of that structure. The standard objection to this defense is that empirical knowledge is only knowledge of experience, not of things as they are in themselves. But this objection has force only if one thinks that what we experience are representations. If one thinks that we experience the world directly through representing it, then our experience will tell us about the world and not just our representations.

The second line of defense is to point out the incoherence of the alternatives. In particular, an internalist view holds that the world has no structure and contains no objects apart from our activity. Yet, almost all our theories imply the existence of objects with structure apart from our activity. (Take, for example, any theory of the origin of the universe or of the planet Earth. If there were no cognizers around to provide the world with sufficient structure to produce objects it is unclear how anything could happen, including the genesis of structures that made those cognizers possible.) This point is so simple it almost seems to be in bad taste to bring it up. Yet simple points such as these, when taken seriously, seem to me to show the incoherence of internalist views. The next section explores the inadequacies of internalist views in more detail.

9.4: Inadequacies of Internalism

In this section I argue for the view of representation and its role in knowledge presented here by pointing out the inadequacies of its major rival, internalism. The arguments I use will be aimed at Putnam's internal realism, but most of them apply to other forms of internalism as well. These criticisms center on the internalist claims that objectivity is a matter of internal constraints, that objects are constructs within representational systems, and that determinate reference of our concepts is possible because our concepts are active in the construction of their objects and have their application or reference fixed in this process. The criticisms fall into two main groups: First, I will argue that internalist views fare no better than externalist views in explaining how determinate reference is fixed, whether to objects or other words, as long as they still view representations as objects or signs. Second, I will argue that internalism, and especially Putnam's internal realism, is self-referentially incoherent. It has trouble explaining both how it can be possible and how it can be true without implying its own falsity. All of these defects of internalism arise because it retains the very model of representation whose inadequacies led to the abandonment of external models of objectivity. It still sees representations as objects or symbols or other static entities even though it sees that it is impossible for such things to represent apart from our activity.

Internalism attempts to solve the problem of reference by making the things referred to a part of the representational system, so that a veil of ideas does not make it impossible to fix reference. "Since the objects and the signs are alike internal to the scheme of description, it is possible to say what matches what." (Putnam 1981, p. 52) Reference becomes fixed by us by assigning our signs referents within our representational system through our interpretation of the signs. This solution seems plausible only as long as one doesn't attempt to get clear about what a scheme of description is and how reference is fixed within such a scheme.

Putnam diagnoses the problem about reference to external objects in this way:

As we have seen, the problem is this: there are these objects out there. Here is the mind/brain carrying on its thinking/computing. How do the thinker's symbols (or those of his mind/brain) get into a unique correspondence with objects and sets of objects out there. (Putnam 1981, p. 51)

If one substitutes "other symbols" for "objects" in the above quote, the problem remains. For how a determinate reference relation is determined between signs and other signs or objects within the system is just as much a problem as it was for external objects. For the problem with representing external objects was not due to the fact that they were external; it was due to the fact that no object or symbol intrinsically represents any other object apart from situation in an act of interpretation. The veil of ideas did not cause a problem for representation; a problem with our view of representation caused the veil of ideas.

One can be comforted by the assurance that the other symbols or objects are internal to the scheme of description only if one does not worry too much about what such a scheme is. Such schemes can only be sets of symbols, i.e., sets of physical objects or events that can be taken to stand for something else. How it is that unique reference relations are determined between signs and objects remains a problem whether the objects in question are external noumena or other symbols within the system.

Simply being internal to the same representational systems does not establish any unique relationship between items; it only allows us to assign or make connections between items in our interpretations. Our interpretation of the signs does not establish any determinate relation between them either. For we could always make alternative interpretations that satisfied all the internal, formal constraints of the system. To see that this is so, take any of Putnam's model theoretic arguments and let the alternative models consist of signs within the system rather than external objects. There are no built in semantic relations between objects whether the objects are cows, cats, mats, linguistic symbols, or constructed objects in a conceptual schemes. Of course we can assign a determinate reference in an act of interpretation, but the determinacy of the reference lies in the act of interpretation, not in the object. The object itself is indifferent as to how it is interpreted. Within the constraints imposed by the formal structure of the symbol system, the interpretation is wholly a matter of convention. The determinacy of reference lasts only as long as the act of interpretation; the act does not impart any special properties to the object that allow it to refer determinately in the future.

An example of the application of model theoretic arguments to symbol systems is the indeterminacy of translation. Different symbol systems can be mapped onto each other in different ways while retaining the isomorphism required by their formal structures. This is true of subsets of a single system as well; there is indeterminacy of semantic relations within single languages as well. The fact that within a system we can agree by convention on an interpretation does not determine a unique reference relation between the symbols of the system. This only shows that an act of interpretation has determinate reference, not that the symbol system does. There can always be alternative interpretations of the system.

Putnam thought he could solve the problem of reference by doing away with the veil of ideas between representation and object. But the problem did not lie in the veil of ideas, but in the view of representations as physical objects. If the representation is taken to be a physical object then, even when we do have access to both object and representation, no determinate reference is established. This is exactly what Putnam's use of the Lowenheim-Skolem Theorem shows: that even in cases where there can be some relation between sign and object, there will be too many alternative relationships. Thus even in cases where we can determinately interpret our symbols, there will be too many different alternative interpretations. The moral of the model theoretic argument should have been that representations are acts rather than objects. Instead Putnam concludes that if only objects were internal to our representational systems we could make symbols come alive by picking out a preferred interpretation for them. But on internal grounds alone, there is nothing but convention to make one interpretation preferable to another. This is what the Lowenheim-Skolem Theorem shows whether the models are taken to consist of external objects or other symbols.

Another inadequacy of internalism is that it does not avoid the problem that it set out to avoid: the problem of the relation of representations to extra-representational reality. The way that internalism avoids charges of total relativism and idealism is to retain the notion of an external world that somehow constrains our knowledge. In Putnam, this constraint takes the form of experiential "input". (Putnam 1981, p. 54) There is no determinate relationship between any particular representation and the world. Instead the world has input into and constrains the entire representational system as a whole. It is no more easy, however, to see how an entire representational system has a determinate relation to the world in which it can be constrained than it is in the case of a particular representation. Internalism, by making the basic unit of meaning the whole conceptual system rather than the particular representation, does not solve the problem of the relation of representations to the external world. It simply puts the problem off. Since, on the internalist view, there is no semantic relation between representations and the external world it is not clear how interaction with this world can constrain the representations.

The problem of how the external world can provide constraints to our knowledge without there being a determinate relationship between our representations and this external world becomes even more acute when one considers how it is that we construct objects or cut the world up by the application of our concepts. This process is left completely obscure by Putnam. It is difficult to understand how concepts are formed from experience since there is no determinate relation between the concepts and the world nor any intrinsic structure to the world which can constrain their formation.[4] It is even more difficult to understand what guides the application of particular concepts to particular situations. If there is no structure to the world apart from the activity of our concepts then why cannot any concept be applied in any situation? And if there is no determinate reference relationship between concepts and the world then what determines their application in particular situations? The cornerstone of any internalist theory is the claim that objects are constructed by the activity of concepts upon unstructured reality, but it seems impossible to give an intelligible account of how this could occur without ascribing some structure to reality and allowing some determinate reference relation between this structure and our concepts.

The seeming impossibility of making clear such a central tenet of internalism suggests that there may be an internal incoherency involved in the view. For the only way the view could possibly work is if it is false and there is some determinate relation between the world and our representations that allows the world to constrain our representations. Internalism recognizes that the traditional view of representation cannot work, yet it retains a view of the role of representation in knowledge that can only make sense if the traditional model of representation works.

An examination of the status of the subject and the symbol system itself makes the incoherence involved in internalism clearer. All objects are constructs within conceptual schemes, according to internalism. Yet there are two classes of objects of which this cannot be true: subjects and conceptual schemes. Can a conceptual scheme be a construct within itself? The subject that does the cognizing and the conceptual scheme must be given some transcendental status in order for the theory to work, just as the transcendental ego and the categories had to have special status for Kant. The image that Putnam uses for his view betrays this incoherence. He compares his version of internalism to a play in which the author is also a character in the play. (Putnam 1976, p. 138) While this is an intriguing and powerful image, if one tries to take it seriously one finds that it doesn't make sense. A subject cannot be both the maker and the thing made, for the construction of objects is a structured activity, one that requires concepts, and these concepts cannot themselves be the result of the making activity which they structure. All construction of objects requires the application of pre-existing concepts. Internalism cannot be true of the concepts that make internalism itself work. It only seems to do away with ready-made objects; in fact it requires a special class of ready-made objects to become intelligible.

This type of self-referential incoherence can also be seen in the internal realist definition of truth. The central notion of the internal realist theory of truth is idealized justification. The intuitions that lead to the idealization of justification as the definition of truth are inconsistent with internalism. The only reason one would believe that under ideal epistemic conditions conceptual systems will converge is if one believed that under ideal conditions conceptual systems are constrained by a common external world. What other reason could one possibly offer for such a belief?. Putnam has gone through great pains to show that alternative theories can satisfy all the operational constraints imposed by the world. Why should interaction with the world under ideal conditions cause different conceptual schemes to converge, unless there is some determinate relationship between the conceptual scheme and the world under these ideal conditions which constrains the scheme in the interaction? It turns out that the only reason one could have to believe that internal realism is true would be if it was false, at least under ideal conditions. .

Alvin Plantinga has devised a graphic way of showing the incoherence of internal realism. He has shown (Plantiga 1982) that internal realism implies that there exists an ideal rational scientific community. Putnam explicitly denies that there could exist such a community. (Putnam 1981, p. 55; Putnam 1983, p. 84) Plantiga provides a fairly involved proof of this implication of internal realism. I will give a shorter account which illustrates the method of his proof.

Internal realism can be defined in this way:

IR- a sentence X is true iff. if there were an ideal rational scientific community (IRS) it would accept X.

Let X in the definition above be the sentence "An IRS does not exist." If one assumes this sentence to be true, they can arrive at a contradiction using IR. An IRS could not accept this sentence, for doing so would make it true, in which case the IRS would not exist. By a reductio ad absurdum, then, there must be an IRS.

What this shows is that in order to be made intelligible as a theory of truth, internal realism must assume ideal epistemic conditions. It must assume, as an idealization, conditions under which correspondence of representations with reality is possible. In this way, internal realism is true to its perspectivist roots. Perspectivist models of objectivity seek a God's eye view, a representation free from the distortions introduced by the medium of representation. In identifying objectivity with the God's eye view, the perspectivist model makes all other particular representations merely subjective, incapable of corresponding with reality. Internal realism takes this to the extreme. It is simply the identification of true representation with the God's eye view, or the non-perspectival perspective, along with the proviso that the God's eye view is impossible.

This proviso ultimately makes the view unintelligible as a theory of truth. For the type of idealization involved is not like that in science. It still makes sense to speak of friction even if frictionless planes do not exist, but it is not clear that it makes sense to talk of justification if ideal conditions do not exist. If, in fact, we can never arrive at conditions under which our representations can correspond to reality, then talk of justification of our assertions does not make sense. For, as Putnam points out (Putnam 1983, p. xiv), justification and assertion require standards of correctness, or else they collapse into mere noise making. Standards of correctness as idealized limits that are not really possible cannot play this role. As we saw, justification and the internal realist theory of truth, only make sense if there is reason to believe that justification procedures will cause convergence. If the conditions that make convergence possible are only ideals, not possible in reality, then justification cannot even make sense. Justification does not make sense without standards of correctness or truth, but internal realism attempts to define truth in terms of justification.

It is not as if internal justification and representation under ideal conditions were on a continuum of which truth was the limit. It is not clear how any epistemic conditions can be more or less ideal for internal justification. If justification is a matter of the internal characteristics of a system, then the conditions under which the system is applied to external reality should not be relevant to justification. Ideal epistemic conditions are not a natural extension of internal justification standards; ideal conditions for justification only make sense if justification is not entirely internal. Again, internal realism only makes sense if it is false.

Internalism, then, is incoherent both in that it requires the existence of subjects and conceptual schemes which have a ready-made structure and in that it defines truth and objectivity in terms of conditions that are ruled out as impossible by its model of representation. The first type of incoherence above is suggestive of another type of argument against internalism. This type of argument also reveals much about the motivations of realism, especially of realisms in which particular perspectives are seen as providing objective knowledge. I will close with an account of how the view of representation presented here is motivated by these type of considerations.

9.5: The Objectivity of the Self and the Motivation of Realism

One of the most convincing arguments against internalism springs from the nature of our knowledge of self. We all know of at least one object that is not a construct within a conceptual scheme: ourselves. Other people form conceptions of us, yet we know that our nature is not dependent on these conceptions. We are also aware that even our own conceptualizations of ourselves do not exhaust our nature. There are many aspects of ourselves of which we are not cognitively aware, and we can come to learn about these in various ways.

Of course, the way in which we conceive of ourselves is affected by other people's conception of us and by our past conception of ourselves. The affect that this has on what we are cannot be overestimated. What others think of us is one of the most powerful considerations in human motivation. Yet, this does not make us constructs within other people's conceptual schemes. None of us consider ourselves to be such internal objects, for we all attribute a reality to ourselves that is independent of the conceptualization of others. To use Thomas Nagel's terms, there is something that it feels like to be us at any particular moment, and this is not a product of any conceptual scheme, for it can be conceptualized in various ways with more or less success.

The various representations that other people make of us and that we make of ourselves do not exhaust us. The complex network of embodied potentialities that is us, by its very nature as a potentiality, outruns any set of its expressions. The sense that we have of the felt character of the teleological tendencies or potentials that operate in us makes us aware that there is more of us than meets the eye, whether it be the eye of another person or our mind's eye. This is true of all things we characterize as dispositions. Their nature is not exhausted by our representations of them; to paraphrase Shakespeare, there are more things in this world, Horatio, than are dreamed of in your representations. It is this conviction that the world outruns our representation of it that forms the heart of realism.

I am also able to refer to this object apart from the correctness of my representation of it. This, of course, is true of my ability to refer to me. Even if I know nothing about myself (say I am an amnesiac), I will have little trouble in picking out which person in a room is me, or in distinguishing myself from the furniture.

Thus our knowledge of the object with which we are most familiar, ourselves, reveals an object that does not fit the internalist model. It reveals an object with a nature that is determinate apart from external conceptualization and to which we are constantly referring apart from representations.

This response to internalism seems especially powerful to me because it works upon what seems to me to be the strongest of the motivations for realism. It is often said that realism is motivated by a need for firm foundations for our knowledge or for stability. These motivations cannot be applied to intuitive realism, for it insists upon radical incompleteness of our knowledge and on the relativity of the discovery of objects to modes of interaction with the world. On this view we cannot mirror the world with our representations; we can only know it through its causal effects on us. The motivation for intuitive realism springs from the importance of particular perspectives and particular interactions with the world in all of our lives.

Anti-realisms make objectivity dependent on traits or modes of conceptualization that are common to a community, culture, or species. The real world becomes a common world, one shared by all rational creatures in common, or by all members of a community or culture. The private, the peculiar, and the particular all become less than real; they become merely subjective aspects of reality.

For example, the way that the Huron river looks to me as I float on my back on an a mid-summer's night, with the stars reflected in the surface of the water and the dark background of trees mounting up to the sky so full of fire flies that it looks like the heavens have merged with the earth, can find no place in an objective conception of the world on this view. Another person might describe the scene as a rinky-dink stream with lots of bugs and no place to get a pastrami sandwich. If objectivity is a matter of those aspects of our conceptual schemes that are shared by some set of cognizers, then our idiosyncratic views of the world become merely subjective, less than real. Only if particular perspectives can be felt as objective, if we can feel the force of the world determining our experience through its phenomenological feel, can we regard our particular impressions of the world as objective and real.

The world that is important to us, the one whose reality we become excited in arguing over, is not any social construct, rational framework of experience, or model within a theory. One rarely sees vehement argument about the transcendent reality of space and time or about the objective validity of the categories. Yet arguments about realism are often animated. The main issue concerning realism is not the existence of a world apart from our experience; almost everyone believes in that. Nor does it concern the existence of abstract, totally objective aspects of reality; these are too boring to excite much interest in most people. The issue is about the objective reality of our particular experiences of reality.

The world that the realist is vehement about saving is not the noumenal world that no one experiences, nor is it the completely objective world that everyone experiences. They are interested in maintaining the reality and objectivity of their world, their particular experiences of the world.

The realist is rebelling against an internalist model in which the only representations of the world that are objective are those that can be shared by many perspectives or that are common to a whole community. Realists, egotists that they are, are only minimally interested in these bleached out abstract versions of the world. They are interested in their particular experiences of the world. It is my particular experiences of myself, my family, my friends, and of particular places and events of significance to me whose objectivity and reality I want realism to preserve. It is no surprise that realists sometimes take anti-realist arguments personally. These arguments threaten the objective reality of the only source of value in our lives: our particular experiences of ourselves and the world around us. We want the value we get from particular experiences to reflect the world we live in and not just its subjective appearance to us.[5]

Likewise, the realist in science (as opposed to the scientific realist) is also interested mainly in preserving the objectivity and centrality of particular experiences of the world rather than in the reality of some objective constructs or theories. It is the reality of facts that the realist in science fights for, not the reality of theories.

Thus, the real motivation of realism is a striving for what was impossible on the physical-visual model of representation; it is a striving for the objectivity of particular representations of the world from particular perspectives. We want the reality of our personal worlds preserved.

This was impossible when representations were seen as objects that could not determine their own relation to the world. When representing is seen as an active interaction with the world, particular representations can carry within them the possibility of their justification. They can be objective, since the connections made within them can be caused by the interaction with the world rather than by our activity. According to this type of realism, particular representations from particular perspectives can be objective; they can reflect the world with which they interact.

This is fortunate for us, since all of our experience is from particular perspectives. Our self is defined by the particular perspectives it takes toward the world. This type of realism leaves room for the objectivity of this self. We are lucky that it does, for otherwise we would not be characters in a play of our own making, as Putnam suggests. There is no room for particular selves in a world constructed according to the internalist model; only what is accessible from many perspectives has objective reality on that view. If realism were not true, we would be lost characters in search of a play which could accommodate our particularity.