The verification theory of meaning put forward by Moritz Schlick in 1932 lost its appeal in the following years. Yet the basic concept that meaning somehow referred only to experience still remained the fundamental assumption of the Positivists. In an article. written in 1950, Hempel attempted to clear up the problems caused by both the verificationist and the refutationist theories of meaning. The fundamental Positivist tenet of meaning, Hempel said, is that non-analytic knowledge is based on experience. The basic concept of the logical Positivists of his time was that a sentence makes a cognitively meaningful assertion and thus can be said to be either true or false only under the following conditions.

1. It is analytic or self-contradictory.

2. It is capable at least in principle of experiential test.

It is the problem of testability. therefore. that determines whether a statement has empirical significance or not. To explain this Hempel introduced three new concepts.

An observable Characteristic; under suitable circumstances its presence or absence can be ascertained through direct observation.

An observable predicate; a term which designates an observable characteristic.

An observation sentence; any sentence which correctly or incorrectly asserts one or more specific objects have or lack some specified observable characteristic.

We must understand that according to the accepted criteria of the verification theory of meaning, an observation statement asserts something that is in principle ascertainable by direct observation. There is no assertion being made that some person could ever actually observe the condition existing. In order for this to occur, for there to be a complete verification in principle, a sentence has empirical meaning if and only if it is not analytic and follows logically from some finite and logically consistent class of observation sentences. This means that the verification theory of meaning rules out all sentences of universal form and therefore all statements that express general laws.

Another problem with the verification requirement is that given a sentence "S" which satisfies the proposed criteria and sentence "N" which has no empirical meaning. the disjunction "SvN" could be inferred and therefore would have empirical meaning. However, this is exactly what the verification is designed to counter. Thus it makes the verification requirement too inclusive.

On the other hand, given observation predicate "P", the observation sentence which states that at least one thing has the property P. is completely verifiable. However, if we look at its denial; nothing has the property "P", it is not verifiable. Therefore, under the criteria above, the sentence has no cognitively significant meaning. This results in the dilemma that either we must give up the fundamental logical principle that if a sentence is true or false its denial is false or true, or we must deny that "(x) P(x)" is logically equivalent to the negation of "(Ex) P(x)". Therefore, Hempel stated, we must abandon the verification criteria shown above.

Through a similar logical argument Hempel showed that The theory of falsifiability in principle can be shown to have the same kinds of problems and therefore also must be set aside. Ayer. he said, proposed that the rule that a sentence will have empirical import if in conjunction with suitable subsidiary hypotheses it is possible to derive observation sentences which are not derivable from the hypotheses alone. This, it turns out. is too liberal.

One alternative to the verification theory of meaning would require a second language, one in which the problems shown using standard language could not occur. Given this empirical language a sentence would have cognitive meaning if and only if it is translatable to some such language. An empiricist language would have the following qualifications.

1. Its vocabulary contained the customary locutions of logic. certain observation predicates, and all expressions definable in terms of these.

2. The rules of sentence formation would be those laid down in a contemporary logical system such as "Principia Mathematica".

The advantage in such a language is that the problems noted above could not occur. For example the disjunction SvN would be impossible because N would not be a sentence in this new language. Also. the negation of a sentence in this new language would also be a sentence. But this alternative has its own problems. It requires that every sentence to be cognitively meaningful must be translatable into the new language. This means that all of its empirical terms must be definable by means of the allowed observation predicates. Many scientific terms are not so definable which would mean that we would have to reject all scientific hypotheses containing such terms.

In conclusion Hempel offers the following criteria for cognitively meaningful sentences. a sentence has cognitive meaning if its non-logical constituents refer, directly or in certain specified indirect ways, to observables. It does not state what the meaning of a cognitively significant statement is. It neither says nor implies that the meaning can be exhaustively characterized by what the totality of possible tests would reveal in terms of observable phenomena.


In "Verification and Experience, Ayer discussed some of the problems associated with a theory of meaning called "coherence" theory. A theory based on the concept that statements derived their meaning from their mutual compatibility. In the early years of the logical Positivist school science was scoring tremendous successes. The idea that scientific method was the only true way toward truth helped to develop the enormous energy that sustained the Logical Positivists and the Vienna Circle. "Verification and Experience" was written in 1936 and showed that even as early as that some sense of balance was developing in the group. The problem with any Positivist approach to philosophy is that it not only assumes that there is a method whereby we can check propositions against experience through simple observable characteristics, but that it also assumes that was the way that we actually do experience things. I will set aside the criticism because that is not the purpose of this work. I just wanted to introduce the idea that the development of logical positivism was an evolutionary process.

The truth or falsehood of empirical propositions is determined by their agreement with reality. No other Positivist stated the relationship between propositions and reality this simply. The problem most Positivists tried to solve dealt with the relationship between a proposition and some protocol statement that was based on the experience of some person. But, there is no common sentiment on what we mean by agreement with reality. Ayer placed empirical propositions into two groups; propositions, or those whose truth or falsehood was determined by other propositions, and propositions whose truth or falsehood was determined directly by observation. The former include all universal propositions. To test the validity of a universal proposition it is necessary to test the truth or falsehood of a singular proposition referring to a particular member of the universal class. In this case if we find that the proposition is verified we cannot imply from it that the universal is. On the other hand, if we find a case where the singular case is falsified then it in turn falsifies the universal law. He called this logical asymmetry.

What he proposed to call basic propositions are those propositions that can be directly confronted by facts. As long as the distinctions between them and other propositions were unambiguous he felt we could confine our study of empirical justification to the problems of these basic propositions. The difference between these and later concepts of protocol sentences as put forward ny Neurath was that these protocol sentences concern only an individual called by name and the direct observable experience by him of some observation characteristic. Ayer can find no reason for confining them to this restricted view. This led him to a revision of what is generally termed "coherence theory."

The coherence theory of truth is not concerned with the definition of truth or falsehood, only with the means by which they are determined. According to this theory a proposition is to be accepted or rejected according to whether it does or does not accord with previously accepted propositions. But, what if a proposition should not in fact be compatible with other accepted propositions? Why should we reject it? Why not throw out the whole lot and begin again with a new set? However, according to the accepted coherence theory there is only one completely coherent system of propositions. If this were actually true of course, then we would certainly have a reason to throw out the new incompatible theory. However, how can we know that there is just one completely coherent system of propositions? Why couldn't there be more than one completely coherent systems of propositions which are incompatible with each other?

The answer given, Ayer said, is that there cannot be more than one such coherent system if the systems are based on protocol Propositions. It is a historical fact that while some people occasionally offer incompatible protocol propositions, they are drowned out by the majority. The ultimate answer is to trust to the "scientists of our era". The point is that even given our greatest respect for the scientists of our era, we can hardly accept that the notion of agreement with reality depends only on their behavior. Carnap, he said. suggested that the problems concerning the nature of propositions depend only on conventions with words for their solution. Ayer's conclusion is that the form of basic propositions depend only partly on linguistic conventions. They depend just as well on the nature of the given.