Quine’s theory of meaning

I. Introduction

Before confirming or falsifying a proposition, we first need to be clear on the proposition to see if it is even meaningful. The way we interpret the meaning of the affect the way we go about an epistemic investigation.

Moritz Schlick, the famous German philosopher and one of the founders of logical positivism[1] provides us with a way to understand the meaning of sentences, namely through inspection of its logical verifiability. Quine, on the other hand, thinks that this theory of meaning and verification cannot be successfully established. Quine’s objection to Schlick’s theory of meaning bears direct consequences on the development of science, for if we cannot verify and answer scientific inquires the way Schlick instructed us to, we may have to reexamine our entire knowledge system and turn to an alternative epistemological approach, e.g. holism.

In the following section, I shall start by explaining the arguments and positions of Moritz Schlick and Willard Quine. Though Quine never directly responded to Schlick’s theory, I shall speculate on a potential response in the third section, based on Quine’s direct response to Carnap, another representative of logical positivism. Eventually, I shall extract Quine’s own position on the theory of meaning, and defend it against Steven Wagner’s attack.

II. Moritz Schlick on Meaningfulness and Confirmation

Schlick claims that when presented with a sentence, it should not be necessary for us to ask about what proposition it stands for, unless we have not actually understood the meaning of that sentence. For example, a sentence such as “take me to a country where the sky is three times as blue as in England” has no meaning, and we should not be mistaken into think it does even though it is grammatical. This leads to Schlick’s definition of statements of meaning: “stating the meaning of a sentence amounts to stating the rules according to which the sentence is to be used, and this is the same as stating the way in which it can be verified or falsified. The meaning of a proposition is the method of its verification”. Hence, to give meaning to a proposition is to give the rules of its verification. To drive this point home, Schlick gives us the example of asking what it means when we speak of simultaneity of two events. The answer to the meaning of that sentence consists in a description of experimental method we use to deem two events simultaneous. For example, we may say that the meaning of simultaneity of event A and B is that at a point midway between them in space we would receive at the same time the light ray sent out from both locations at a given time. We cannot obtain the meaning of simultaneity independently of an experimental method, i.e. the method of verification.

So how is verifiability defined? Verifiability is defined as the possibility of verification. Schlick carefully differentiates verifiability from verification, so that we do not fall into what Schlick and C.I.Lewis calls “here and now predicament[2]”.

We then go into the categorization of “possibility” in the definition of verifiability, which can be of two kinds: empirical and logical. Schlick chooses the latter: verifiability is the logical possibility of verification. To understand why, let us examine both kinds. Empirical possibility means possibility under the limitation of natural laws. However, we do not have or even potentially cannot have a complete understanding of the natural laws at the current stage. Had we adopted the meaning of verifiability to be the empirical possibility of verification, we would largely and undesirably limit ourselves to a small set of meaningful statements. For example, we cannot verify a statement such as “the back of the moon is unilluminated right now”, since we are physically restricted [3]to the extent that it is empirically impossible for us to verify this statement. However, we are clearly reluctant to give up this statement or deem it meaningless. Let along the fact that our ability to physically verify is improving all the time, but it does not make sense to wait for a statement to become meaningful simply because at some point we’d be able to physically verify it.

On the other hand, a statement is logically possible if it obeys the rules of grammar we have stipulated. A logical impossibility[4] is an inconsistency. A logical impossibility occurs not because our imagination cannot reach them[5], nor is it because that there is no representational correspondence of it in the real world, but because the definitions of the words are incompatible with logic. If we adopt logical possibility, then there would be no ambiguity as to what can be verified.

Hence, we can distill Schlick’s theory of meaning to the following: verifiability is the logical possibility of a verification. And meanings are equated with conditions of verification.

One of the consequences of Schlick’s theory is a specific method to carry out scientific inquiries: we examine an individual claim to ascertain its verifiability. Then we investigate by scientific methods. One of the examples Schlick used to demonstrate this line of scientific inquiry is immortality[6]. Immortality had always been as a “metaphysical problem”. However, since it is logically verifiable, it is merely an empirical hypothesis rather than metaphysical. The significance of this categorization is that it saves immortality from being an impossible problem, because if it is just an empirically meaningful hypothesis, there are possible experiences that could falsify or confirm immortality. Though it seems like an impossible problem to us right now, it is of same type as all the other impossible hypotheses---their impossibility is merely of the empirical kind instead of logical.

III. Willard Van Orman Quine on the First Dogma of Empiricism

In the famous Two dogmas of empiricism, Quine criticizes the dogma of reductionism, which is the position that each meaningful statement derives its meaningfulness from certain logical constructions linked with immediate experience. To fully understand where this criticism originates from, we need to examine Quine’s attack on the first dogma---the distinction between analytic and synthetic statement.

A statement is analytic when “it is true by virtue of meaning and independently of fact” (21, Quine. Quine first emphasizes that meaning is not the same as reference, since we generally understand “morning star” and “evening star” to have different meanings. We then run into difficulty trying to define precisely the definition of meaning. Hence Quine turned away from an analysis of meaning and sought alternative explanations.

The first approach to explaining analyticity is synonymy. This is a worthwhile strategy because we may be able to conclude analyticity without any analysis of meaning if we can simply By substituting parts of a sentence with their synonyms[7], we can preserve logical truth. But that begs the question of where synonymy comes from. The first attempt we may take to explain synonymy is by looking at the definition of words. However, we cannot turn to a lexicographer for an answer, since the synonymous relation between “unmarried man” and “bachelor” are established merely based on how these words are used. We can imagine that over time, our habits of language evolve, and the dictionary definitions would also change. Analyticity would also change if we appeal to dictionary definitions for truth conditions of analytic propositions. In other words, we have appealed to empirical facts while maintaining analyticity. This already defeats the distinction between analytic and synthetic. Quines admits that there is a type of synonymy that does not rest on prior experience, namely the new notations introduced for the sake of abbreviation. This type of synonymy is created directly by definition. But this type of synonymy also does not hold water once we examine the role of definition formally[8].

The second attempt we make for synonymy is by appealing to interchangeability between words while holding fix the truth value of the sentence. We may want to say two words are synonymous if by interchanging them, we preserve the truth value of the proposition (call this “interchangeable salva veritate”). However, “bachelor” and “unmarried man” are not always interchangeable. An example given by Quine is that we cannot perform the substitution on the sentence “’Bachelor’ has less than ten letters”. To avoid obvious counter example as such, Quine proposes that we examine only “cognitive synonymy”. However, how can we get a cognitive synonym not presupposing analyticity? Here, the type of interchangeability we require is the type which ensures that necessarily all bachelors are unmarried men. For the word “necessarily” to be part of the language sensibly, it is required that the language, it is supposed that we have already made satisfactory sense of “analytic”. Our argument is therefore circular. In other words, we cannot find a satisfactory account of why necessarily all bachelors are unmarried man.

One may argue that even though we cannot agree on a definition of analyticity, it does not mean we should throw out the distinction all together. Is Quine denying the usefulness of distinguishing the palm from the back of the hand simply because there is not a clear division where our palm starts?

In Quine’s defense, I think it should be noted that this lack of clear division between the analytic and synthetic has a different nature from our lack of division between the back of our hand and our palm.

Clearly, even if we do not know where exactly our palm starts, it does not mean that there is not a difference between palm and back-of-hand. However, the lack of distinction between analytic and synthetic may be due to the circular nature of analyticity itself. Quine is not trying to show how we cannot categorize a statement, but any categorization should require us to understand the nature of the category. Quine has shown that the nature of analyticity is ambiguous. The palm of the hand is the opposite. There may be times where we cannot tell whether the side of our hand is the front or the back, but never have we ever doubted that we know exactly what a palm is.

Additionally, there is a test for palm or back of the hand. If I point at somewhere on your hand, you will be able to output whether it is the palm or the back. We can always draw a line dividing palm and back arbitrarily, and the distinction would still be useful. But we are not willing to draw any arbitrary distinction between analytic and synthetic. We are convinced that it is a inherent property that a statement is analytic. But in that case, our search for such an inherent property has failed, which casts doubt on whether there is anything inherent at all. If there is none and the distinction between analytic and synthetic has to be arbitrary to a certain extent, then we might as well not have such a distinction. Therefore, the analytic and synthetic distinction, according to Quine, is a “metaphysical article of faith”.

Quine’s attack on dogma two lands on the verification theory and reductionism. The verification theory says that statements are synonymous “if and only if they are alike in point of method of empirical confirmation or information”. It hence provides us with a way to define synonymy, which can serve as a support for the analytic synthetic distinction. But Quine has already suggested that that distinction cannot be saved, hence there must be something wrong with verification theory. But Quine allows us to examine the verification theory itself for its flaw, instead of appealing to the doomed analyticity.

Let us examine the nature of the relationship between a statement and the experience which confirms it. The first option is what Quine calls “direct report”, which is a form of radical reductionism[9]. If we take radical reductionist point of view, we also must take the entire statement as a compounded unit to translate into our sense data, instead of individual terms. This means when we verify, we give statement a priority over terms. Quine then takes on the analysis of Rudolf Carnap’s reduction, since Carnap was the only one who not only agreed with radical reductionism, but also actually carried out the reduction. The problem that Carnap and any radical reductionist suffers, is that it was not indicated how exactly a statement can be translated into physical experience. For example, the statement “Quality q is at (x, y, z, t)” cannot be translated into sense data and logic, because “is at” is an undefined connective in our language of sense data. Problems as such indicates deeper flaws of reductionism, which assumes that each synthetic statement corresponds to a unique range of sensory events that can dictates whether the statement is of a truth value. At last, Quine seals the coffin of reductionism by saying that as long as we assume that the truth value of a proposition is in its confirmation, it would presuppose that there is a subset of statements that are vacuously confirmed, leading us back to the distinction between analytic and synthetic statements. The dogma of reductionism is fundamentally the same as the dogma of analyticity.

IV. On the Consequence of Rejecting the Second Dogma

Even though Quine did not explicitly respond to Schlick, we can fill in the response for Quine. Let us take the example [10]used by Schlick, “The lady wore a dark red dress which was bright green”. Shlick would say that this sentence is meaningless, not because of how our mind works or anything empirical that we observe, but simply and merely because it represents a logical impossibility. What would Quine say to respond? We know that Quine would not agree with Shlick’s analysis. Even though Quine admits that logical assertions are analytic, he does not consider it to be a useful test for natural languages. Quine would however agree that this sentence is false. Its falsity comes from our practical definitions of dark red and bright green--- the definitions are incompatible. Here, the term “definition” may be slightly overloaded, since I am not only referring to the vernacular or pictorial definitions of color, but rather a snapshot of a comprehensive scientific understanding of what colors are, what wavelengths they occupy, how they propagate through our visual system and give us the perception of themselves. Quine would say that given all these empirical facts that we have collected on physics, neuroscience and cognitive science, we can conclude that it is false to say that a dress cannot be dark red and bright green at the same time without appealing to verificationism.

It may seem at the first glance that Schlick has anticipated such a response from Quine by ruling out the empirical possibility account of verifiability. Empirical evidence is important for Quine, but Quine does not need to appeal to empirical possibility. We need to carefully distinguish between Quine’s view on verification from just an application of empirical possibility. Let us take the example Schlick used to refute empirically possible verifiability: “What is the other side of the moon like right now?”. It seems that it is empirically impossible for us to answer that question. But the approach that Quine would take is to bring in a system of knowledge we have about the moon, rather than appealing to any immediate empirical possibility. For example, we already have 3D modelling of the moon, we already think of induction as an useful empirical principle in science [11]which means that the other side of the moon would not change as we ask the question, and what we already know would let us safely conclude what the other side of the moon looks like.

Schlick may further argue that there are statements that would seem like nonsense to us until we have developed appropriate empirical toolkit to examine it in the future, but we should not need to wait to conclude on meaningfulness. But does Quine require us to be able to verify a statement empirically immediately? I imagine that for Quine, a holistic approach also would not limit us to only verify what has already been known to us here and now. For example, though Schlick may say that we would not be able to verify E = 📷until we actually had the derivation that led to this result, Quine can respond that a statement as such, though not directly confirmed by our knowledge back then, would never the less have always been consistent with the network of knowledge we had[12]. Hence, the fact that it fits perfectly into our network of knowledge is an indication that it can just be another construct of ours which we call scientific theorem [13], which can serve further purposes for us.

Quine thinks that we experience the world as a whole. I have tried putting this thought into use when anticipating a discourse between Quine and Schlick above. This leads to Quine’s proposal of holism, which says that the unit of empirical significance is the whole of science, and that every component in a proposition has its meaning contingent on the entire network of knowledge and belief. This view can be seen as the opposite of Schlick’s position which sees the meaning of each proposition as independent from any other and entirely made up of the meaning of every component. [How does this affect Schlick’s position]

V. Net of knowledge

Before diving into Quine’s holism, let us first examine a metaphor constantly used by Quine: our system of knowledge is a net. With this metaphor in place, he then distinguishes claims “in the center of the net”, and empirical claims that are “on the peripheral”. If we understand the metaphor bluntly, we have to specify exactly how the net is weaved, and what exactly “the peripheral” corresponds to. Using the metaphor bluntly is troublesome. I propose that we reformat the metaphor to use a directed graph instead of a “net”. In this directed graph, each theory or lemma corresponds to a node. An edge is placed between node u and v and goes from u to v if theory u implies theory v. For example, let lemma v be “apples fall from the tree”. Then a node u that points to it would be a theory of gravity.

With this new metaphor, what Quine called the “center” of the net corresponds to nodes with high in degrees, and the “peripheral” are the nodes that have lower in degrees. A node in this graph can have a high connectivity without necessarily being called the “center” of the graph. Namely, all its edges can be out-edges. Intuitively, that means taking down such a node from the net does not have that much implication on the rest of the net.

One may object to this correspondence by pointing out the difficulty of “high” and “low”. However, note that in all contexts where these notions are used, they are used in a relative sense rather than absolute. In other words, it is unnecessary to define a threshold for high connectivity. All we need is the ability to pick out the nodes with relatively high connectivity out of a set of nodes.

Another potential objection may be that there are uncountable number of theory and observations even in one person’s daily life. However, for any information to be efficiently represented in us, we have to categorize, summarize, and condense. No matter how many observations we make, we can only distill a countable number of them into expressible theories. Same analogy applies to color: though there are countless number of colors, we can only name a few. I can show you a thousand shades of red, but in your head all of them are represented as a category “red”.

VI. Quine on the Theory of Meaning

Let us extract here on Quine’s theory of meaning, which can be summarized as the theory of holism. How can we tell that a proposition is meaningful? Quine says that we ought to turn to empiricism without these dogmas, and that approach would give us a test of meaning not for individual propositions, but rather for our fabric of beliefs as a whole. Our knowledge system is a net with experience determining how the edges adjust, and experience is only impinged on the edges. So to test whether a belief in the net is meaningful is to test whether a belief is consistent with the beliefs interweaved with it in the net; to test whether a belief on the peripheral of the net (or in other words, to expand the net) is meaningful is to test whether it is also consistent with experience. This may sounds somewhat abstract, but let us use the example in the last section, where we demonstrated how we know an expression such as “a dark red and bright green dress” is false, we see that it is an expression that does not fit into our net in the sense that it is not compatible with what we know about the colors “dark red” and “bright green”.

Even though our system of beliefs and meaningful propositions form a structure similar to a belief net, let us not be mistaken into thinking that this net is a set-and-stoned absolute golden standard for the judgment of truth. For Schlick, the expression “dark red and bright green dress” is a logical contradiction; however, for Quine, this does not represent a logical contradiction, but an empirical contradiction only in the current system. An empirical contradiction is a proposition which is only contextually false. Given a different context or system of belief, it may be no longer contradictory to say something is red and green at the same time. Nothing is inherent in the status of red and green that is logically incompatible with each other, because otherwise “something cannot be red and green at the same time” can just be reduced to an analysis of the meaning of “red” and “green”. It is not illogical to say that a dress is red and green at the same time. However, it is unempirical to say so, because every piece of empirical evidence we have tells us otherwise. In other words, there could exist a coherent system of knowledge and beliefs in which, for example, due to different compositions of human retina and perception of light, we see a dress as red and green at the same time.

And what if in this world we actually one day come across a red and green dress? We should not abandon this observation. Instead, we should reevaluate all statements interconnected with it. According to Quine, “re-evaluation of some statements entails re-evaluation of others, because of their logical interconnections”, and no proposition is immune to revision. This leads to the questions of how we select the statements to revise. Quine gives us two guiding principles to follow. The first principle is that the germaneness of experience to certain propositions should reflect the likelihood with which we choose [14]one proposition over the other. The second principle is that we should disturb the system as little as possible. This would usually mean that we prefer to revise peripheral propositions rather the central ones, keeping the knowledge net intact. Though Quine does not explain why we would have the second simplicity principle, we can fill it in for him, assuming that conventional scientific practice is not the only reason.

Lots of philosophers would say that we cannot know for certain that a theory possesses the status of truth metaphysically, so I am not making a claim that simplicity of changes directly entails metaphysical truth. Rather, I argue that probabilistically, the more central claims have a higher chance of being true. When two contradictory theories (call them A and B) are present, one whose being changed would bring down lots of other theories, and the other which could just be a singleton error, we tend to think the fault is at theory B. The intuition is based on the fact that the previous claim has been tested a lot more than the other new theory, if we see every theory which is based on theory A and is functionally predictive as a test. Hence, by making as minimal changes to the system as possible, we are preserving truth maximally probabilistically.

Hence, to summarize, Quine’s theory of meaning is a theory of holism. The meaning of an individual proposition is interpreted and tested on the basis of the entire system, and the falsification of an individual proposition can never be done fully independently without evaluating further implications it has on the system. He has not only provided us the method of expanding our net of knowledge (expanding the peripheral based on experience), but also the methods of maintaining and updating the net (the two principles).

VII. Objections and Response

In Quine’s Holism, Steven J. Wagner wrote:

One can see the failed attempts to move from a view of evidence that recognizes embedding[15]--- holism in one sense--- to a rejection of distinctive verification conditions for statements--- holism in the other.

What Wagner points out here is an inconsistency in Quine’s holism, which is that if we are to treat Quine’s holism as a method of verification, then we are still distinctively verifying individual statements---the exact consequence that holism tries to get away from[16]. Wagner admits that Quine says when we question one hypothesis, we simultaneously question any that imply it, so that no hypothesis is ever tested alone. But he thinks that is irrelevant, and the question is whether intuitively distinct hypothesis have different verification conditions.

If we recognize the concept of embedding [17]individual propositions into a net of knowledge, then we have to recognize the individuality of the statements. No matter how we verify a statement, we are still verifying a single statement.

I think Wagner’s mistake of thinking of this as an inconsistency lies in a conception of what Quine is trying to achieve. Quine’s goal is not to choose a system of claims that are inherently truer [18]than the other, but rather one that is more useful. It is an equivocation of “verification” to say that Quine rejects the “verification” of single statements, since by confirming or infirming a statement, Quine is making no judgment about its metaphysical status at all, unlike Shlick. Shlick’s verification means that we can confidently say that the statement is metaphysically meaningful and can be true. For Quine, “verification” simply means accepting or rejecting a statement into our knowledge net. In that sense, in order to build up the knowledge net, Quine cannot nullify all acceptance conditions of individual statements. Now looking back at the inconsistency Wagner tries to point out, we see that the first sense of holism is actually consistent with the second part if we rephrase it as such: to move from a view of evidence that recognizes epistemological embedding, to a rejection of metaphysical verification of individual statements. The two senses of holism reconcile in both seeking a stable and useful system of knowledge pragmatically. There is no inconsistency in Quine’s system after all.

VIII. Conclusion

In this paper, I have tried to present a conflict between Quine and logical positivists such as Schlick on the issue of theory of meaning. I have tried to present Quine’s theory of meaning, which is a theory of holism by reforming his analogy of a knowledge “net”. Future research should be devoted into investigating what a holistic theory of meaning entails for the fields of science.

Acknowledgment

I want to thank Professor Teddy Seidenfeld who guided me through the initial stage of selecting a topic of my interest and general directions to which I should expand my paper upon. I also appreciate the interesting and comprehensive lectures that he gave on Shlick’s logical positivism. Without Professor Seidenfeld, it would have been a lot harder to finish this paper.

Reference

Quine, W. V. “Main Trends in Recent Philosophy: Two Dogmas of Empiricism.” The Philosophical Review60, no. 1 (1951): 20. https://doi.org/10.2307/2181906.

Schlick, Moritz. “Meaning and Verification.” The Philosophical Review45, no. 4 (1936): 339. https://doi.org/10.2307/2180487.

Wagner, Steven J. “Quine's Holism.” Oxford Journals, January 1986, 1–6. https://www.jstor.org/stable/3328733.

[1] a form of positivism is a philosophical system that holds that every rationally justifiable empirical assertion can be scientifically verified or else admits a logical or mathematical proof). It was developed by members of the Vienna Circle, which considers that the only meaningful philosophical problems are those which can be solved by logical analysis.

[2] “suppose it maintained that no issue is meaningful unless it can be put to the test of decisive verification. And no verification can take place except in the immediately present experience of the subject. Then nothing can be meant except what is actually present in the experience in which that meaning is entertained”. This possesses a problem because…

[3] Physically restricted in the sense that we are not able to instantly send a detection device to verify our statement.

[4] One of the examples of logical impossibility given by Schlick is “the lady wore a dark red dress which was bright green”.

[5] In fact, Schlick is very strict on separating meaning from any psychological representation.

[6] This is not the question of living forever, but rather the question of “survival” after death.

[7] For example, given the sentence “No bachelor is married”, we can substitute “bachelor” with “unmarried man”, and get “No unmarried man is married”, which can be considered to be a logical truth, further implying the analyticity of the original sentence.

[8] (explain Quine’s position on definition)

[9] Radical reductionism formally states that every meaningful statement is held to be translatable into another statement about immediate experience.

[10]This example was used to illustrate how a sentence is meaningful if and only if it is logically possibly to verify it.

[11] Here, explain how Quine says that scientific principles are nothing but convenient construct. They do not stand for ontological truth.

[12] At least the portion that Einstein has developed, which we can call the more precise portion.

[13] Every scientific theorem for Quine is nothing more than a convenient construct like the irrational numbers, this is further illustrated in the next section.

[14] By “choosing A over B”, I mean grant A truth value and B false. The purpose of choosing is to restore the coherence of our knowledge system.

[15] Wagner defines the “embedding phenomenon” as such: “the relevance of sensory evidence to s depends on the position of s in surrounding theory”. This definition itself suggests a negligence of individuality of propositions.

[16] Quine’s exact words goes as such “No statement, taken in isolation from its fellows, admits confirmation at all. Our statements face the tribunal of sense experience only as a corporate body” (Quine, 38).

[18] Here, “true” refers to the property of a metaphysical truth. [what is a better way to phrase “truth”?]