New Philosophy for a New Millennium

(A series of notes prepared as part of an informal research

 project between 2000 and 2005)

By Mike Holliday


Section 2: 'Mirroring Reality' - How Not to Think about the World

1 The Mirror of Nature

I noted in the Introduction that the major trend in Western thought over the last few hundred years has been its emphasis on rationality. One important aspect of this is that modernist philosophers and scientists, from Descartes and Galileo through to the Enlightenment and beyond, saw themselves as attempting to describe how the world really is. This notion has been the subject of much criticism since the end of the nineteenth century, from Nietzsche, Heidegger, Wittgenstein, Kuhn, Derrida, and many others. One of the more recent critics, writing from a perspective within Anglo-Saxon analytic philosophy, is Richard Rorty, whose book Philosophy and the Mirror of Nature (1979) is a sustained attack on the coherence of the idea that knowledge consists of accurately mirroring in our minds the world outside, an achievement that is possible only through those special mental processes that supposedly form the backbone of the methodology of modern science and of Western philosophy. According to Rorty (1979, p. 12), our philosophy has been dominated by 'Greek ocular metaphors':

The picture which holds traditional philosophy captive is that of the mind as a great mirror, containing various representations - some accurate, some not - and capable of being studied by pure, nonempirical methods. Without the notion of the mind as mirror, the notion of knowledge as accuracy of representation would not have suggested itself. Without this latter notion, the strategy common to Descartes and Kant - getting more accurate representations by inspecting, repairing, and polishing the mirror, so to speak - would not have made sense.

I want to concentrate on the views of another of the critics of the tradition of the 'mirror of nature', namely Hilary Putnam. My main reason for concentrating on Putnam is that his arguments introduce us to two themes that will reappear in various guises at a later stage. The first of these is that there is no single 'correct' description of reality; and the second is that our objectifications of things as entities, separated out from the rest of reality, depend on our cognitive processes, which in turn arise from our interactions with the world. I will therefore not be primarily concerned with ascertaining whether Putnam's arguments are technically correct or not - to an extent I am taking the broad thrust of his conclusions for granted, for the reasons set out in the Introduction.

2 Putnam's Criticism of Metaphysical Realism

Putnam argues against a stance which he believes has been the default for modern philosophers, and which he terms 'metaphysical realism'. This is a position that Putnam himself once espoused, but which he has subjected to a sustained critique since the mid-1970s. Metaphysical realism is broadly similar to Rorty's 'mirror of nature', and is characterized by the following three theses: (i) that 'the world consists of a fixed totality of mind-independent objects', (ii) that 'there is exactly one true and complete description of the way the world is', and (iii) that 'truth involves some sort of correspondence' (Putnam 1990, p. 30).

It is important to make clear at the outset that Putnam isn't arguing against a position that has been definitively held by any particular philosopher; his point is rather that metaphysical realism represents the underlying assumptions that inform the current approach to analytic philosophy. One can certainly criticize Putnam's arguments by attempting to show that it is possible to adopt a moderate form of realism without fully accepting Putnam's alternative; a good, recent example is Khlentzos 2004a. However, I aim to make use of Putnam's arguments in order to display the difficulties associated with the idea that 'the truth' reflects, or corresponds to, 'reality'; a view that is almost a truism in its everyday sense, but which is problematic when taken as a form of metaphysics. My consideration of metaphysical realism is therefore, firstly, an initial stepping stone enabling me to introduce ideas that I will later explore in more detail and, secondly, aimed at undermining any resistance to those ideas that is based on the view that reality is surely just a reflection of the sum total of our (true) descriptions of the world. 

But what is the relationship between the three tenets that are taken to comprise the metaphysical realist position? Each single claim does not logically necessitate the others, but Putnam believes (1990, pp. 30-31) that there is a natural fit between them. First of all, an obvious way (but not, he admits, the only way) to make sense of how there can be 'exactly one true and complete description of the world' is as follows: if we assume a definite set of individuals that constitute the world, together with a definite set of properties and relations, then a language that contains names for each and every individual, and predicates for each and every property or relation, can provide a true and complete description consisting of the set of true sentences concerning the world. In other words, thesis (i) provides the backdrop to understanding thesis (ii). It is then natural to identify the individuals with the names, and the predicates with the properties and relations. In other words, (i) plus (ii) naturally suggest thesis (iii). Putnam also notes (1983, p. 211) that a correspondence theory of truth requires a 'ready-made world', otherwise any number of different structures might 'copy' the world, and truth would lose its non-perspectival character. In other words, (iii) suggests (i).

Putnam contrasts metaphysical realism with a position that he terms internal realism,1 which holds that the question what objects does the world consist of? can only make sense within a theory or description. This alternative stance implies that there is no 'God's Eye View' that represents true knowledge (Putnam 1981, pp. 49-50). Therefore Putnam's position is not a denial of realism tout court; he emphasizes that there is a reality separate from our own thoughts, but maintains that there is no single, objectively correct, description of that reality from an external point of view.

Putnam has used a number of different arguments to criticize metaphysical realism, of which I want to consider two. The first of these is described as the model-theoretic argument. Secondly, there is Putnam's discussion of conceptual relativity. The model-theoretic argument has occasioned a large volume of commentary in the philosophical literature; the argument from conceptual relativity is, however, equally interesting - indeed Ernest Sosa, in his review (1993) of Putnam's critique of metaphysical realism, suggests that the case based on conceptual relativity is easily the most persuasive.

3 The Model-Theoretic Argument

The term 'model-theoretic' derives from model theory, which is the branch of mathematical philosophy that studies the interpretation of languages (formal or informal) using set-theoretic concepts. We won't need to delve into model theory itself, since it is relatively easy to understand Putnam's model-theoretic argument in an intuitive way, without the theoretical apparatus that surrounds it. The model-theoretic argument is stated somewhat differently by Putnam in three separate texts, but the version that I will concentrate on is that contained in Putnam's book Reason, Truth and History (1981), a version that has been termed 'the permutation argument'.2

Let's start by considering an example of how the permutation version of the argument actually works (taken from Iseda 1997):

The basic idea is that the world consists of five objects (identified by Greek letters). We also have a language that contains the three terms 'electron', 'Nana', and 'cat', together with logical terms such as 'all', 'some', etc. In this language, we can form and then evaluate sentences such as 'Nana is a cat', 'no electron is a cat', and so on. An interpretation is a schema that maps this language to the domain of objects. Interpretation I and Interpretation J map the terms of the language to different objects. But, and this is the critical point, each of the sentences that we can form will have the same truth value under both interpretations, whereas metaphysical realism seems to require that one and only one interpretation should be capable of producing all of the true sentences that we can make about the world. With a large number of objects and predicates, we could obtain as many different interpretations as we like by simply permutating the objects.3

To see more clearly how both interpretations can produce the same set of true sentences, we should note that the Greek letters are simply placeholders (in Iseda's example, 'alpha' is actually the entity 'Nana'). So it makes no difference whether it is 'alpha' that is Nana or 'epsilon' that is Nana. This is even clearer in the diagram below, where the world contains only the five objects (denoted by the Greek letters), but no properties belonging to them; in this case we could not make any statements about the objects, we would only be able to name them ('That is Nana', 'That is Fred', etc.). This makes it clear that the Greek letters identifying the objects have no content and are just placeholders.

Another way of looking at this is to note that in Iseda's example, under both interpretations, (i) all terms in the language have extensions with the same number of items, and (ii) 'Nana' has an extension that is an element of the extension of 'cat'. This reflects the fact that, as Putnam conceives metaphysical realism, it is the structure, not the identity of the elements within the structure, that is supposed to explain the meanings of the terms involved. This brings out the fact that metaphysical realism is a purely syntactical theory. All we require is the correspondence between the name and the object (as indicated in Iseda's example by the vertical lines); there is nothing inherent in 'alpha' or 'epsilon' which has anything to do with whether one of them is 'Nana'. It is this syntactical nature of metaphysical realism that means that Putnam's argument against it can be set out in model-theoretic terms. The argument is couched in formal, axiomatic terms, not because Putnam believes that this is the way in which such arguments should be couched, but because the argument is a reductio ad absurdam of a position that holds that there are two different domains but no 'metaphysical link' between the two, only a structural link.4

This characterization of metaphysical realism as an attempt to link two domains, the domain of our descriptions and the domain of objects, needs to be understood in the context of the overall dialectic that drives Putnam's rejection of metaphysical realism. Putnam states in the Preface to Reason, Truth and History that his main aim is to break the hold of the dichotomy between the objective and subjective views of truth, aiming instead for an approach to truth which will try and unite the objective and the subjective and which might be characterized as 'the mind and the world jointly make up the mind and the world'. Much of the discussion that follows in the first three chapters of Reason, Truth and History is intended to show that there are problems with what might be termed 'traditional' theories of truth and reference, for example that 'meanings are just in the head' and the 'similitude theory'. Putnam suggests that, having rejected such 'traditional' approaches, the most common view in analytic philosophy is that an interpretation of a language is fixed collectively by means of operational and theoretical constraints. The attraction of this is that we could, in principle, tell whether a theory is in fact true, since whether the operational and theoretical constraints are met is an empirical matter. But, says Putnam, this view must be incorrect, because it tries to fix the intensions and extensions of individual terms by fixing truth conditions for whole sentences, and Quine has already shown that this does not work. Putnam then uses the permutation argument to extend Quine's conclusion, such that reference of terms is (radically) underdetermined even if we fix the truth-values of all sentences in all possible worlds.

We can now see that the form of the model-theoretic argument against metaphysical realism is therefore that of a reductio, as Putnam himself (1994a, pp. 280-281) later made clear:

I argued [in Reason, Truth and History] that metaphysical realism leaves us with no intelligible way to refute ontological relativity, and concluded that metaphysical realism is wrong. And I still see ontological relativity as a refutation of any philosophical position that leads to it. ... I am happy that most of the readers of Reason, Truth and History (though regrettably not all) understood that my argument was meant as a reductio.

In chapter three of Reason, Truth and History, Putnam adds some comments that shed further light on the nature of the difficulty faced by the advocate of metaphysical realism. He points out that we would have no difficulty accounting for how our descriptions can refer to entities in the external world if we were to assume that there existed 'self-identifying objects' or some other form of 'magical theory of reference', but goes on to claim that modern-day philosophers would find such entities unacceptable. This gives metaphysical realists a particular problem because they conceive of language and of objects in the world as two separate realms, and are therefore obliged to give some account of the link between them. However, the only remaining candidate seems to be some form of structural correspondence between the two domains, but it is difficult to provide a coherent account of how such a correspondence can be identified without falling back on objects that somehow identify themselves as matched to our terms of reference. For example, how (on the metaphysical realist's account) can we use the word 'cat' to refer to all, and only all, objects that are cats, unless those objects identify themselves to us as cats? Therefore metaphysical realists are put into the position of wanting to 'think of the world as consisting of objects that are at one and the same time mind-independent and Self-Identifying. This is what one cannot do' (Putnam 1981, p. 54).

Putnam provides another way of describing this difficulty: the separation of the two domains means that external reality seems to retreat to a version of Kant's noumenal world, yet it doesn't seem possible to identify the required mapping between our concepts and objects in the noumenal world unless we have direct access to the noumenal objects themselves. '[The model-theoretic argument] simply states in mathematical language the intuitive fact that to single out a correspondence between two domains one needs some independent access to both domains' (Putnam 1981, pp. 73-74).

4 Summing-Up the Model-Theoretic Argument

Up to this point, the model-theoretic argument is generally regarded as uncontroversial; the results from model-theory and the permutation argument are accepted as being correct. The concern of critics has been whether they tell against the realist. And here most discussion has centered on whether the realist can determine reference by identifying one interpretation of a language as being the 'intended' interpretation. An obvious realist response to Putnam would be that reference is determined by something other than operational and theoretical constraints; the most common approach being to use a causal theory of reference to fix the link between words and objects. Putnam's discussion of this response contains what has been termed his just more theory argument, which can be stated quite briefly: '... a "causal" theory of reference is not ... of any help here: for how "causes" can uniquely refer is as much of a puzzle as how "cat" can, on the metaphysical realist picture' (1976, p. 126). In other words, unless you hold that there is some direct and mysterious grasp of the relationship between a sign and that to which it refers, you have to provide some explanation as to how that relationship is singled out (or else your theory seems unintelligible). But simply referring to that relationship as a 'causal' relationship doesn't help because (on the metaphysical realist picture) it presumes the ability to identify a unique, intended relationship (as opposed to some other relationship, e.g. causal*). The discussion of this point in Putnam 1981 is along similar lines:

To me, believing that some correspondence intrinsically just is reference (not as a result of our operational and theoretical constraints, or our intentions, but as an ultimate metaphysical fact) amounts to a magical theory of reference. ... Even if one is willing to contemplate such unexplainable metaphysical facts, the epistemological problems that accompany such a metaphysical view seem insuperable (p. 47).

This element of the debate has generated more discussion than any other aspect of Putnam's critique of metaphysical realism. The detailed issues are not particularly germane to my own interests, and so I will not pursue them further here; however, a summary of the arguments can be found in the notes.5

After much discussion in the literature, it seems that an impasse has been reached on the model-theoretic argument, and specifically the just more theory issue, with both Putnam and his critics arguing that the other side's premises beg the question (Sosa 1993). Hale and Wright (1997) conclude that, although Putnam's model-theoretic argument may not be conclusive, he does show that the metaphysical realist faces severe problems in setting out a coherent case for reference being determined by whole-sentence semantics, and that the alternative of a causal theory of reference looks pretty hopeless:

That there are correct and incorrect things to say about what expressions refer to is enough for there to be truths - at least on the conception of truth favoured by the internal realist - about reference. ... What, precisely, might be put in doubt by [Putnam's arguments] is the existence of truths about reference in a more substantial sense of 'truth', a concept of truth whose applicability to claims of a certain kind requires, beyond the unimpeachability of those claims in the light of the ordinary discipline that informs their use, some form of robust fit between them and the world.

My prognosis is broadly the same as Hale and Wright's.  It may be difficult to make Putnam's argument against metaphysical realism watertight, but he has shown there are significant problems with the claim that reference and truth are determined by the external world, independently of our conceptions, and therefore with the view that that there is a single correct description of reality towards which we are impelled by the world itself.

5 The Argument for Conceptual Relativity

Now let's move on to the second of Putnam's arguments against metaphysical realism, the argument based on the idea of conceptual relativity. This is usually expressed by Putnam by means of one or more of a series of examples, such as the following:6

(i) The number of objects in a room might be counted by taking either the number of ordinary sized objects (a table, a chair, and a lamp: three objects) or the number of elementary particles. Putnam suggests that it is only our interests that make us take, say, the lamp as one object rather than a particular group of particles. (And why, asks Putnam, is the lamp one object, when we wouldn't consider the lamp and a piece of chewing gum stuck to it as just a single object? Especially as the gum might be difficult to remove whereas the shade might fall off the lamp whenever we move it!)

(ii) Physicists can construe fields (e.g. an electromagnetic field) either as objects or as constructions using particles acting at a distance.

(iii) We can take points as primitive and explain spheres by means of sets of points, or we can explain points from sets of convergent spheres, thereby taking the spheres as primitive. In other words, we can describe space-time by means of points or we can describe space-time using a language that takes points to be simply limits.

(iv) Suppose we have three marbles in a box; how many objects are there? A logician who counts mereological wholes as objects would say 'seven', rather than 'three'.

The idea that the examples are trying to get across is that when we describe a particular situation, that situation does not, of itself, legislate how we use words such as 'exist', 'object', and 'property'. In Sosa's (1993) words:

Each of us acquires and develops a view of things that includes criteria of existence and perdurance for categories of objects. When we consider whether an object of a certain sort exists, the specification of the sort will entail the relevant criteria of existence and perdurance. And when we correctly recognize that an object of that sort does exist, our claim is elliptical for '... exists relative to this our conceptual scheme'.

But if the metaphysical realist were to accept that even such basic properties as existence and cardinality are relative to a conceptual scheme, then the 'real world' as conceived by the metaphysical realist seems to retreat to a totally unknown Kantian noumenal world, with all its attendant epistemological difficulties (Putnam 1976, p. 133). The metaphysical realist cannot accept conceptual relativity because that idea is rooted in the fact that the notions of 'object' and 'existence' have a number of different uses rather than one determinate meaning, and metaphysical realism, if it is to mean anything, must of necessity hold that there is but one meaning of such terms (Putnam 1987, p. 19). Hence, argues Putnam, the fact that conceptual relativity does exist indicates that metaphysical realism is not tenable.

Putnam (1994a, p. 309) suggests that the existence of conceptual relativity indicates not that truth doesn't really depend on things external to us but that the nature of that dependence changes as the conceptual schemes that we use change. It is not that we legislate facts by our use of language, but rather that we can only talk of facts once we have adopted a particular way of discussing matters. Therefore one must not interpret Putnam as claiming that, say, the existence of elephants depends upon the existence of humans and their concepts (Khlentzos 2004b). It is quite clear that we can evaluate a possible world without humans and say 'Yes, elephants exist in such a world', because we evaluate from our world, with our conceptual schemes (Sosa 1993).

I suggest that the picture of one, unchanging dependence between our language and the world only seems sensible if we imagine ourselves as having some sort of passive existence outside the world. Given such an externalist account, it might seem that the only way for there to be a relationship between our words and the world is for objects to be predetermined in the world, and hence for the dependence of our words on those objects to be some form of correspondence relationship. But if we reject that entire externalist viewpoint, and instead consider that we ourselves are part of the world and are to a large degree constituted by our interactions with it, then the rationale for maintaining that what counts as an object is predetermined seems to disappear. This is an alternative way of viewing our relationship with the world that I will develop further in Section 3: An Alternative Metaphor.

6 Putnam's Alternative to Metaphysical Realism

First, however, let's see how Putnam develops his own account of how we refer to the world. What is needed, Putnam believes, is some form of internalist perspective, one that recognizes that we are not outside reality, but are part of it, and hence does not attempt to sustain a description of the whole of reality from an external point of view. Putnam's 'internal realism' holds that the question what objects does the world consist of? can only make sense within a theory or description, and that there is therefore no 'God's Eye View' that provides the one, true description of the world. This is not an outright rejection of a reality separate from our own thoughts, but a positive acceptance that there is no single, objectively correct, description of that reality. The common ground that internal realism shares with other 'realist' positions has been characterized by George Lakoff (1987, p. 158) as: (i) 'a commitment to the existence of a real world, both external to human beings and including the reality of human experience', (ii) 'a link of some sort between human conceptual systems and other aspects of reality', (iii) 'a conception of truth that is not merely based on internal coherence', (iv) 'a commitment to the existence of stable knowledge of the external world', and (v) 'a rejection of the view that "anything goes" - that any conceptual system is as good as any other.'

Putnam (1987, p. 17) characterizes internal realism as the insistence that realism is not inconsistent with conceptual relativity, i.e. with the view that the world does not legislate precisely how we should use words such as 'object' and 'exist'. So he agrees (1999, p. 7) that knowledge claims must be 'responsible to reality' but points out that the forms of knowledge claims and the ways in which they are responsible to reality are not fixed once and for all. He therefore says (1988, p. 115) that the essence of internal realism is that 'truth does not transcend use'. 

One advantage claimed by Putnam for internal realism is that reference to external objects becomes straightforwardly explainable because both signs and objects are internal to the scheme of description. To refer back to the problem of the metaphysical realist, whom Putnam characterizes as being forced to maintain that objects are both self-identifying and mind-independent, we can say that for the internal realist, objects are in some sense self-identifying, since they fall under conceptual schemes that we have constructed. But then they are not mind-independent (Putnam 1981, pp. 53-54).

Putnam acknowledges (1981, pp. 56-64) that his internal realism owes a substantial debt to Kant, who was the first philosopher to seriously criticize the correspondence theory of truth and to attempt to explain objective knowledge without presupposing that there is a way that the world is that is independent of our own conception of the world. Given that Putnam's views owe much to Kant, it would be useful to consider the manner in which internal realism appears to differ significantly from the original Kantian intuitions. The important difference is that Putnam views the adoption of internal realism as implying the rejection of any Kantian notion of noumena or 'things in themselves'. As he explains it (1981, pp. 63-64), Kant saw the 'power' in the noumenal world to cause sensations as ascribable to the noumenal world as a whole. So if we have a sensation of a horse or a chair, this does not, for Kant, imply there must be 'noumenal horses' or 'noumenal chairs'. Indeed, this is why Putnam says that Kant did not hold a correspondence theory of truth. All one can say, according to Putnam's interpretation (1981, pp. 63) of Kant, is that 'any judgment about ... objects ... says that the noumenal world as a whole is such that this is a description that a [being with our rational nature] given the information available to a being with our sense organs ... would construct.' Kant believed that the idea of a 'thing in itself' might be empty of content, but nevertheless retained the formal sense of the concept within his metaphysics. However, Putnam (1987, p. 36) believes that once we realize that we cannot know the things as they are in themselves, and accept that objects and existence are relative to a conceptual scheme, we can see that the whole concept of something 'as it is in itself' is incoherent, and hence entirely reject the notion of a noumenal world and the resulting metaphysical dichotomies.

7 The Existence of 'Facts' about the World

Having run through Putnam's arguments for internal realism, I now want to discuss two types of objection to this position. The first objection takes the view that the type of metaphysics proposed by Putnam effectively denies that there are 'facts' about the world that make our statements true. Certainly, John Searle takes this to be the case in his discussion of 'anti-realist' positions in Chapter 7 of The Construction of Social Reality (1995). The ontological status of 'facts' is, however, debatable. The usual view of facts is that they are those items in the world that make propositions true, but many philosophers have a concern that this notion too closely resembles that of a true proposition to enable it to play the explanatory role that is required of it (Loux 2002, pp. 164-168). 

However, Searle (1995, p. 166) wants to clearly distinguish descriptions from facts and states of affairs, claiming that '... we do not make "worlds"; we make descriptions that the actual world may fit or fail to fit'. He adds that, once a conceptual scheme and set of definitions have been fixed, then the aspects of the world that accord with those definitions have an existence that is independent of the definitions. For Searle (1995, p. 166), facts or states of affairs can only be described relative to a set of linguistic categories; they do not exist relative to a set of linguistic categories. In chapter 9 of The Construction of Social Reality, Searle reconstructs the correspondence theory of truth such that it becomes simply a matter of how we use the words 'fact' and 'correspond' when we say that a statement is true if and only if it corresponds to the facts.

... the whole point of having the notion of 'fact' is to have a notion for that which stands outside the statement but which makes it true, ... if it is true. On this account facts are not complex objects, nor are they linguistic entities; rather, they are conditions, specifically, they are conditions in the world that satisfy the truth conditions expressed by statements (Searle 1995, p. 211).

So 'corresponds to the facts' is not some strange sort of general relationship that must hold between our (true) representations and reality, but simply a term for the various ways in which statements can represent the world (Searle 1995, p. 213). 

What does Putnam say on this issue? First of all, he claims that, even if we accept (as he does) that the sky is blue independent of how we talk, we do not thereby have to agree that there is something called 'the true proposition that the sky is blue' that exists irrespective of whether there are people who think and talk: 'It is statements (not abstract entities called "propositions") that are true or false, and while it is true that the sky would still have been blue even if language users had not evolved, it is not true that true propositions would still have existed. If language users had not evolved, there would still have been a world, but there would not have been any truths' (Putnam 1994a, p. 302). Elsewhere (1994b, p. 247), he points out that saying that two different descriptions are about 'the same state of affairs', or 'the same event', is simply a commonsense description of what we are doing, and does not involve elevating 'state of affairs' or 'events' into some form of universal ontology. And he comments approvingly on Davidson's claim that

we must not think that true sentences correspond one by one to special objects which 'make them true', call them 'states of affairs'; for to do that would be to bloat our ontology with what Collingwood called 'a kind of ghostly double of the grammarian's sentence' (Putnam 1994a, pp. 300-301).

I conclude that there is no substantive difference between Putnam and Searle on this issue, and that there is little prospect of a strong argument against internal realism based on the objection that it denies widely-held claims about the status of 'facts'.

8 The Possibility of Incompatible Views

A second possible objection to an internalist stance is to suggest that it holds that two 'incompatible views' can both be true, and then point out that this is only possible if we deny the basic rules of rational argument, thereby leaving the internalist open to a facile form of relativism that accepts that any view is as good as any other. Let's start with some of the things Putnam says concerning the possibility of 'inconsistent views'. (i) In Reason, Truth and History (1981, p. 73), we have the following: 'If all it takes to make a theory true is abstract correspondence ..., then incompatible theories can be true. To an internalist this is not objectionable: why should there not sometimes be equally coherent but incompatible conceptual schemes which fit our experiential beliefs equally well?' Note that Putnam refers here to incompatible theories and schemes, not incompatible propositions. (ii) In Putnam 1990 (p. 40), he claims that the reason why objects can be said to be theory-dependent is that we can have different theories, each with an ontology that is incompatible with the ontologies of the other theories, yet all theories can still be correct (e.g. for a physicist, fields can be either objects or constructed from particles). Again, he talks about theories and ontologies, rather than propositions. (iii) In Putnam 1994b he says:

Blackburn writes as if I hold that 'genuinely inconsistent propositions' can be true. Of course, I do not hold this. When I said that propositions which are inconsistent from the point of view of classical semantics can be true, I was criticizing 'classical semantics', not endorsing the conclusion Blackburn ascribes to me (p. 244).

As an example, he points out that in a mini-world of three individuals, one wholly red and one wholly black, the statement 'there is an object which is partly red and partly black' is false if our ontology consists only of the individuals, but true if our ontology includes mereological sums. In such cases, suggests Putnam, the propositions concerned should not be conceived of as 'genuinely inconsistent', and it is the different use of terms such as 'object' and 'exist' that blocks any imputation of genuine inconsistency. So Putnam's view is not that there are multiple perspectives that cannot be conjoined; what he is insisting on is that there is no single, privileged ontology from which to derive one, uniquely-true, perspective (1994b, p. 254). Hence there is no genuine inconsistency of propositions, and conceptual relativity, as Putnam conceives it, is not open to the objection that it implies that we have to give up on the idea of a rational approach to knowledge because the logical connectives lose their content (1994b, p. 247).

9 Putnam's Transition from 'Internal Realism' to 'Natural Realism'

In Reason, Truth and History (p. 55), Putnam held that truth was an idealization of our notion of rational acceptability, in the sense that a statement might be said to be 'true' if we consider that it would be justified under epistemically ideal conditions. The rationale behind this approach was that one cannot straightforwardly identify truth with rational acceptability because truth is a property of a statement that cannot be lost, whereas rational acceptability can be lost . However, Putnam later came to see this view as defective. The major difficulty is that, if there are problems as to how we can have referential access to external things, then there is equally a problem as to how we can have access to 'sufficiently good epistemic conditions'. The source of his error, as Putnam (1999, p. 18) now sees it, is that his 'verificationist semantics' still retained the idea of an interface between ourselves and the world. But Putnam had by now realized that the mind is not an organ that is located 'in here', and that we should instead conceive of it in an active, relational sense. If we do this, then we no longer have a problem with explaining 'how the mind hooks up to the world' because our very conception of what the mind is incorporates a direct relationship with the world. Once he saw this, Putnam acknowledged that internal realism still had the 'outside world' providing referential access to 'sufficiently good epistemic conditions', but he had no way of explaining how it did this, given the lack of a direct relationship between mind and world that was implied by his interface conception.7

Despite this change in his views, Putnam (1994b, p. 242) still believes that the notion of truth is intertwined with the notions of rational acceptability and sufficiently good epistemic conditions, and with the way that words are used. He believes that he is still arguing for the same conclusion - that the principles of metaphysical realism are incoherent - but in a different way (1999, note 41 on p. 183). Accordingly, I don't intend to enter into a detailed discussion of Putnam's move from 'internal realism' to what he terms 'natural' or 'commonsense' realism.

However, I do want to refer briefly to one aspect of Putnam's recent writings on the nature of representation, namely the emphasis that he now places on the relevance of perception. Putnam (1994a, p. 281) says that, after writing Reason, Truth and History, he realized that there was a strong link between the problem of reference and the problem of perception, which is that they are both concerned with the relation between thought and the world. After all, says Putnam, if we didn't think there was a philosophical difficulty concerning how we perceive the world, then we would surely have no problem with how we are able to refer to it. The question 'how does language hook onto the world?' cannot seem problematical unless we have already rejected the response 'how can there be a problem talking about trees and houses, when we see them all the time?' (Putnam 1999, p. 12). He goes on to say that adopting a philosophy that consists of 'Cartesianism plus materialism' implies that our cognitive abilities do not extend beyond our own perceptual inputs, and that the link with the 'external world' must therefore be a causal, not a cognitive, link; and it is this that makes the objective references of our terms appear completely undetermined.

This link between perception and reference suggests to Putnam (1994a, pp. 290-291) that part of the reason why it does not make sense to imagine a world where 'cats' refers to cats* rather than to cats, is that it is easy to perceive that something is a cat and not easy to perceive that something is a cat*. Sensory perception is not a passive affectation of some object termed 'the mind' but experiences of the world by a living being. When we talk about the mind, we are actually talking about a set of abilities that depend upon our brains and upon our interactions with the environment (Putnam 1999, pp. 37-38). Once again, this is suggestive of an alternative way of viewing the relationship between our selves and the world, a way that I shall explore in more detail in Section 3.

10 Putnam's Critique of Metaphysical Realism - a Summary

My discussion of Hilary Putnam and his critique of 'metaphysical realism' was designed to highlight two notions that will feature in the following sections. The first of these is that our objectification of things as entities, separated out from the rest of reality, depends upon our cognitive processes, which in turn arise from our interactions with the world. An intuition along these lines is in fact widely shared by researchers working in a variety of fields. Here are a number of comments by philosophers, psychologists, and cognitive scientists that reflect the notion that the world is not simply divided up into definite objects that then make themselves available to our perceptual and cognitive processes:

a) The psychologist Nick Humphrey (1992, pp. 16-17) considers that both subjective experience and the objective world are historical creations that emerged during the evolution of life. He says that, before the emergence of sentient beings,

the phenomena that we now call the phenomena of the material world were not yet in existence: ... because no one was there, there was not ... anything that counted as a volcano, or a dust-storm and so on. I am not suggesting that the world had no substance to it whatsoever. We might say, perhaps, that it consisted of 'worldstuff'. But the properties of this worldstuff had yet to be represented by a mind.

b) Stan Franklin (1995, p. 300), an artificial intelligence researcher, notes that 'Things don't come with bar codes so that I can tell what I'm sitting on is a chair. Neither do categories such as water, clouds, calculus class, quarks.'

c) Next, a succinct statement from the phenomenologist Maria Villela-Petit (1999, p. 513): 'A stone or a galaxy does not open any world. It is not for a stone that there is another stone; it is only for a living being that there can be stones or any such objects ... .'

d) The philosopher Andy Clark (1997, p. 50) puts it this way: '... a human perceives a chair as "affording sitting," but the affordances presented by a chair to a hamster would be radically different. Perception, construed this way, is, from the outset, geared to tracking possibilities for action.'

I shall be arguing that we need to take the intuition that lies behind these comments seriously (to borrow a phrase used by the philosopher David Chalmers).

The second notion that I wanted to introduce is the rejection of the idea that there is a single 'correct' description of reality. Such an idea is the product of two things: (i) our naive realism concerning the world about us, transmuted into a metaphysical thesis, and (ii) the dualism that has permeated Western thought since Descartes, with two domains (variously characterized as matter/objective/noumenal on the one hand, and as soul/subjective/phenomenal on the other hand), but little in the way of a coherent method of linking the two other than the use of merely illustrative phrases to the effect that the latter domain 'reflects', or 'represents', or 'corresponds to' the former domain. As Richard Rorty points out, this myth, or (better) this metaphor, underlies much of Western theoretical thinking over the last few hundred years, with great practical consequences. But myths and metaphors are not to be rejected out of hand; they are in fact vital to our attempts to understand what would otherwise be a completely mysterious universe - without the ability to utilize our intuitive understanding of one realm in the consideration of realms of which we have no intuitive understanding, we would have no chance whatsoever of understanding the latter. But if the metaphor of 'the mirror of nature' has outrun its usefulness, where do we look for an alternative that might be more fruitful in terms of the insights that we can derive, and in terms of its fit with other aspects of human life? Section 3 discusses such an alternative metaphor …

Mike Holliday (July 2005)

Here are links to:

To send me an email, click here.


[1]  Putnam has changed and refined his views over the last 30 years, and accepts that the term 'internal realism' has usually been applied to whatever position he has held since he first coined the phrase in his paper Realism and Reason in 1976. He has also said that the connotations of the word 'internal' are unfortunate, which is why he has sometimes used the phrase 'pragmatic realism' (see Putnam 1994a, note 23 on p. lxvi).

[2] The different versions of the model-theoretic argument are as follows:

(i) Realism and Reason (Putnam 1976). This version is stated quite briefly and uses basic theorems from model theory.

(ii) Models and Reality (Putnam 1977). This is an expanded version of the approach in Realism and Reason, explicitly employing the Lowenheim-Skolem theorems from model theory.

(iii) Reason, Truth and History (Putnam 1981), especially pp. 32-38 and the Appendix. This version has been termed 'the permutation argument' by Hale and Wright (1997), among others. Hale & Wright give two reasons why they concentrate on the permutation variant of the model-theoretic argument: (i) it contains the additional arguments needed to close off various possible ways that a metaphysical realist might evade Putnam's conclusion; (ii) the model-theoretic resources used are much more modest (Hale & Wright 1997, pp. 428-429). Hallett (1994) comments that the permutation version is preferable because it is 'simpler and more general'.

[3] Douven (1997, p3) gives a useful example showing how a series of different interpretations can be straightforwardly produced using the permutation technique.

[4] As Lakoff (1987, pp. 230-231; see also pp. 252-253) puts it, 'Model theory is ... the natural mathematization of objectivist semantics. What Putnam is suggesting is that there can be no such possible mathematicization. That is, objectivist semantics cannot be made precise without contradiction.' This is also why, at a crucial stage in the argument, Putnam seems to just assume that 'true' equates with 'true on some interpretation'; this is simply what 'truth' is taken to be in model theory (Lakoff 1987, p. 232). The only comment that can be made here by a metaphysical realist is that, although the statements that we can make might well be true on some interpretation, it may not be the intended interpretation. Putnam's response is that, given an interpretation which meets all our operational and theoretical constraints, it is difficult to see what additional constraint there could be on reference that picks out the 'intended' interpretation, and that the supposition that an interpretation which meets all the operational and theoretical constraints might really be false seems to 'collapse into unintelligibility' (see Putnam 1976, p. 126).

[5] The main point that is made against Putnam's 'just more theory' argument is that it incorporates a 'use-mention' error, and hence begs the question against a causal theory of reference. Hale and Wright (1997) characterize Putnam's supposed error as follows: 

The onus legitimately placed upon [the metaphysical realist] is not to demonstrate that determinate reference is possible, but to provide a constitutive account which explains how determinate reference works. Accordingly, he is perfectly within his rights to assume, at least pro tem, a metalanguage in which a determinate account of the putative mechanics can in principle be given.

Before looking at Putnam's response to these criticisms, we need to acknowledge that Hale and Wright are correct, and that Putnam did indeed take metaphysical realism and internal realism to be providing explanations as to why true sentences have the property of being true, rather than giving us definitions of the property of 'being true' (which is taken in its ordinary sense) (Stoutland 2002). Both positions are therefore 'substantive-explanatory conceptions of truth'. Accordingly, one of Putnam's starting points is the contention that the correspondence theory of truth must be empirically based, in the sense that the putative correspondence relationship must be capable of being evaluated in terms of some set of operational and theoretical constraints. Therefore his exposition of the just more theory argument assumes that any resort by a metaphysical realist to a causal theory of reference would be to an explanatory theory, rather than (i) an identification of reference with a particular form of causal relationship, or (ii) a position that relies on some form of 'direct intuition' or 'self-identification' of the relevant relationship. As Stoutland explains:

... the kind of causation needed for a causal theory of reference is causal explanation, which is inescapably interest-relative and irreducibly intentional, and hence relative to a language. To argue that there is a sense of 'cause' which escapes this would be to assume that 'cause' is a term whose reference may be fixed absolutely, which would beg the question by assuming what the argument from a causal theory of reference was supposed to establish, namely that there is such a thing as the reference relation—that is, a distinction between a correct and incorrect way of assigning reference (Stoutland 2002, my emphasis).

Now let's turn to the responses that Putnam has made to criticism of the just more theory argument. Firstly, in Putnam 1983 he says that causal realists often claim that he caricatures their position, and that they do not claim that reference is fixed by the conceptual connection between the terms 'reference' and 'causation' but by causation itself. He then adds:

Here the [causal realist] is ignoring his own epistemological position. He is philosophizing as if naive realism were true of him ... What he calls 'causation' really is causation, and of course there is a fixed, somehow singled-out, correspondence between the word and one definite relation in his case. Or so he assumes. But how this can be so was just the question at issue. ... to think that a sign-relation is built into nature is to revert to medieval essentialism, to the idea that there are 'self-identifying objects' and 'species' out there (1983, pp. xi-xii).

The point that Putnam is making here is that the causal theory of reference is either (i) an empirical theory, in which case it is caught by the model-theoretic argument, or (ii) it is a theory of direct (or 'magical') reference, in which case it doesn't explain what we wanted explaining, i.e. how signs can refer to a mind-independent world. 

Secondly, in Putnam 1990 (pp. 82-83) he sets out the issue as follows: (i) the metaphysical realist needs something to constrain the 'true' interpretation to one intended interpretation, (ii) the epistemological situation is such that we have no way of knowing whether the constraint is actually met (due to model-theoretic considerations), (iii) so the metaphysical realist argues that it is not 'our mind' that fixes on the correct reference relationship, but the world itself (i.e. the constraint is out there in the world), (iv) but this idea that the world has a semantic element to it is contrary to our modern-day physicalist understanding, (v) so you cannot be a metaphysical realist and a physicalist. Putnam goes on to say (p. 85; see also note 9 on p. 328) that what he finds unintelligible about the causal theory of reference is the idea that one relation is intrinsically capable of fixing reference all by itself, without any supposed 'metaphysical glue'.

Thirdly, let's consider once again Putnam's dialectic in the first three chapters of Reason, Truth and History. As I interpret Putnam's discussion here, the underlying problem with metaphysical realism is the separation of the two domains of words and objects. This separation coexists with a purported link between the domains, a link which is supposed to constitute reference and which is therefore usually held to determine meaning and truth. But the separation of the domains leads to an obvious and troubling question, namely 'how are meaning and truth possible?', to which there seem to be a small number of alternative approaches. Firstly, one can hold that meaning and truth reside wholly in the domain of words, in which case the separation of the domains is irrelevant. This is the stance to which Putnam's 'Twin Earth' argument is directed (1981, pp. 18-19), with the conclusion that 'meanings aren't just in the head'. Secondly, one can suppose that there just is a link between the two domains. But simply stating that there is such a link between the separate domains doesn't appear to help us much. Here we seem to have two choices: either we can say that the link is of a specific type, e.g. a causal link, or we can think of the link's existence as some sort of basic metaphysical fact that is a part of the nature of the universe. If we choose the first alternative, then in the view of Putnam, we are thinking of the relationship as lying in the domain of objects. For example, he says (1983, p. 207) that any specific correspondence relationship 'being a relation to things which are external and mind-independent, is itself something outside the mind, something "external"!' But if the causal link itself lies in the objects domain, then our problems with reference reoccur; presumably, we can try and identify the existence of the link by means of operational and theoretical constraints, but now this reference is radically underdetermined. This is, of course, precisely where the 'just more theory' argument occurs in Putnam's dialectic.

Rejecting a causal link leaves us with the second of the two choices, namely that reference is a brute, metaphysical fact. Putnam clearly rejects this position on the grounds that (i) it is a reversion to an outdated ontology, and (ii) it does not provide any explanation of reference. If we choose this alternative, we don't have a problem of reference as such, but we have major epistemological problems and seem to have adopted what Putnam terms a 'magical theory of reference'. In fact, Putnam doesn't believe that any modern philosopher will actually adopt this position. He says (1981, p. 51) that modern realists have a problem because they want a correspondence theory of truth, but being materialists they find the notion of self-identifying objects (or other forms of magical reference) objectionable. Ultimately, the modern day metaphysical realist seems to be forced back to mapping between concepts and things, which Putnam has shown leaves reference underdetermined.

[6] See Putnam 1987, pp. 17-21, 32-36; Putnam 1988, pp. 110-116; Putnam 1990, pp. 96-104; Putnam 1992, pp. 115-123.

[7] For an informative account of Putnam's transition from 'internal realism' to 'natural realism', see Maitra 2003, especially pp. 66-71.