In The Limits of Realism, Button provides his version of the model- theoretical argument which was previously discussed by Putnam, and challenges the external realist on whether constraining reference is possible. He offers his version of just more theory (JMT) maneuver and argues that any statement of constraint with empirical content is just more theory. But if the constraint does not have empirical content, it will lead to what he calls Kantian angst, which suggests that we have to accept the skepticism that we cannot rationally infer that we know anything or whether our language is meaningful, which leads to a contradiction. In this essay, I will argue from an external realist’s point of view by providing a possible way of fixing references and denying that external realism will either lead to the Just More Theory maneuver or Kantian angst. 
Button first offers the model- theoretical argument. It suggests that for an ideal theory we can consider to be empirically adequate, it has to “imply predictions which seem to be true” and there is no “one” way of connecting language and the world. There is, however, an extremely large number of ways of connecting language and the world, and the ideal theory can be made true in many ways. In that case, no words can refer to any empirical content in particular, or as Button says,  “an utterly uninterpreted theory is just a bunch of strings of symbols”. This problem is what the Just More Theory (JMT) maneuver refers to. In order for language to carry meaning in a fixed way, it requires a constraint fix on reference C, so that the language and the component of the world can be correspondent. Button starts by offering a few different versions of constraint that contain empirical content, and finally argues that the JMT maneuver is extremely general and can apply to any constraint no matter what the constraints are. If constraint C contains empirical content, there are many other interpretations of what the constraint actually is, in which the constraint can be made true in any way, so that the constraint can not have fixed meaning. This leads to the conclusion that constraint C has to have no empirical content. But if the constraint is unobservable or has no empirical content, we have to accept the skeptical worry of what Button calls Kantian angst- the worry that our language is meaningless or that we can not use language to represent anything. And the conclusion to which it eventually leads–the acceptance of Kantian skepticism–seems contradictory and unacceptable for anyone who is using language to convey meaning. Thus external realism is contradictory and unacceptable. 
As he points out in his essay, Button suggests that the model-theoretical argument and JMT maneuver turn the Cartesian worry(*1) into the Kantian worry. In other words, from the worry that what we think is true might not be true, to the question, how is thinking of the truth even possible?  It would eventually lead to the question, how is correspondence between language and reality even possible? Or, how is the constraint on reference possible at all?
In this paper, I will provide a possible solution for how the constraint on reference could be possible in an ideal theory without running into JMT maneuver or the Kantian angst. I will also offer a counter argument from an anti- external realist’s view, and respond back from an external realist’s perspective (*2).
In my view, whether the constraint fix on reference has or does not have empirical content is independent of the discussion of whether the constraint is functional or not. The determining factor relies on whether or not the constraint is functional, instead of whether the statement of C can successfully refer to specific empirical content in the right way (JMT) or the worry that it does not mean anything. My solution of how to make C functional is through the agreement between the communicating parties. In other words, what makes the correspondence principle functional is some degree of agreement or translation. 
Let’s consider the following situation. In future ideal theory T, there is the constraint C that is fixing the language and reference. According to Button’s argument, there are C1, C2, C3... that are equally true and organized compared to C but with different permutations. So the question becomes, in which way can we say C is the right way of fixing reference and make the theory true, considering the existence of C1, C2, C3…?
From my perspective, the degree of truth of C and C1, C2 and C3… is the same. But the reason why we can say that C can successfully represent the true relationship of the sign and reference is based on the agreement of who is using it, or who is communicating. In other words, as long as all parties during the communication and the context of representing the truth agree with the uses of C, instead of C1, C2, C3… We can successfully make C functional and say that C can successfully represent the world. In that way, it does not matter whether C contains empirical content or not. If it is containing empirical content, C will also apply to C itself, so that C is not just more theory that can still carry meaning and be functional. And if C does not have empirical content, it does not necessarily lead to Kantian skepticism, because the agreement of all parties on the use of C secures the meaning of language, and we can be sure that our language is not meaningless. In other words, it is conceivable for a situation that the constraint C in the future idea theory does not posit natural properties which makes it more true than other competing constraints, but simply the agreement of all parties who are communicating allow C to successfully convey the truth.
This scenario suggests that language can successfully carry meaning and refer to the objective reality while fulfilling the corresponding principle without running into the JMT maneuver and Kantian angst. 
As a counter argument, Button can simply argue that if the agreement has any empirical content, the statement of the agreement between each party is just more theory. In order for this argument to be valid, however, he has to assume that it is not likely for any communication to function without language or the fixing of reference and C cannot apply to the agreement itself. However, as a rational being who can successfully communicate with other rational beings, we can infer that communication is possible without a written agreement of C. It doesn’t need to be written in the language prior to the success of conveying meaning. For example, two people can point out to the constraint C and nod their heads (or any other way that we can successfully communicate, and we can test whether or not it is successful by simply understanding what other people mean). In other words, as long as we can make any successful communication, the constraint C would be functional. In that way, due to the testability of understanding, it would be arbitrary to even consider, for example, an action of pointing out the same constraint and nodding heads, to be just more theory or worry that it does not mean anything. 
From Button’s point of view, he can argue that according to the Cartesian principle, there is always a veil that is between how the world appears and how it actually is. Even if the whole society or all parties agree on one standard of constraining, it is conceivable that in the unobservable realm, there is some veil that is permuting language and reference, so that the whole society believe that they are successfully referring to the same thing in the realm of unobservable but they are actually referring to a very different thing. Therefore, it is impossible for society to have an agreement on all of the relationships of language and to what it refers.
From my view, according to the very same Cartesian principle, external realism is not obliged to believe that the correspondence principle should also be functional in the realm of the unobservable. An external realist can posit the view that it is in theory possible to use language to represent only the empirical world, and the realm of the empirical world is truth evaluable. In other words, the fact that social agreement can only function within the scope of observable does not violate both the Cartesian principle and Correspondence principle. As long as it is possible to create a link between objective reality and language in the realm of observable, the correspondence principle survives. 
Anti-external realists could argue that the truth that depends on a societal agreement is only truth relative to certain societies or parties. They can argue that the truth in specific contexts cannot be true for those who are not using the same constraints. In other words, it is conceivable that the truth that is accepted in a certain group of parties will no longer be true in other parties who decide to use constraint C1. But truth should be objective to all parties despite which language or constraint that they use, and thus external realists cannot claim that the ideal theory can successfully represent the truth.
From my view, the trueness of the theory does not reside on the constraints or the language it uses. For the future ideal theory, as long as there are translations between C1 and C2 on which the parties agree,  they can both accept the trueness of other theories. This means, from an external realist’s point of view, an ideal theory T1 with the constraint fix on reference C1, and an adequate theory T2 with the constraint fix on reference C2, is equally true, and that they are in theory translatable. And the translatability (in principle) makes the ideal theory true despite whichever adequate constraint it chose. To give an example from our ordinary language, we use the word “cat” to refer to the empirical information what we collectively group together as cat, and the word “animals” to refer to the categories of animals. It is in principle possible to use a symbol 1 to refer to cats, and another symbol A to refer to animals, and a sign “-” refer to  “is” or “are”. When we say 1-A, it means that “cats are animals”. The JMT maneuver suggests that there are infinite amounts of signs such as 2, 3, 4... can all individually refer to  “cats”', and B, C, D… can all individually refer to animals. However, in my view, 1-A, 2-B, 3-C are equally true as long as there is a functional constraint C that is fixing the reference to the physical reality. In order for the constraint to be functional, it needs to have an agreement on which constraint they are using. For constraint C that is fixing the sign “1” to cats and “A” to animals, it is in principle possible to translate into C1- which is fixing the sign “2” to cats and “B” to animals.  
In other words, there are two elements that will make an ideal theory true: 1. The successful representation of the physical world- the objective empirical information (*3) that an agent can receive in principle. 2. The true representation of the objective logical relationship in physical reality (*4). This means, as long as the mapping of the “physical world” is complete and organized enough, no matter what set of signs is referring to which sets of information (“objects”), they all carry the same meaning and convey the same logical relationship. That is, it is the successful representation of the empirical information and the correct logical relationship between them that makes the theory true.

*1:Cartesian principle suggests that even an ideal theory might be false, or that the reality might be drastically different than what it entails.
*2:For the purpose of this essay, I am not making any argument toward the existing system of language or current theory. But rather, providing a possible scenario for the future ideal theory, and to show the possibilities that it does not necessarily lead to contradiction or an absurd conclusion like Button suggests. 
*3: According to modern physics, information that can be received “objectively” is not completely the same for different observers in theory, which might be a challenge for the realist. However, “objective reality” can be considered 1. as an approximation- due to the similarity of the information that we assume each observer shared based on the similar structural way of processing information. 2. as the stable mathematical relationship between observable information, and that relationship can be considered as “objective” without approximation.
*4:Anti-external realists can argue that realists will face the problem of how we can know that there is such “objective physical reality” to which we can collectively refer. My reply is that as long as the communication of “objective physical reality” is possible, the realist can safely assume that without worrying too much due to the Cartesian principle.

Button, T. (2015). The limits of realism. Oxford Univ. Press.
Back to Top