In the early to mid Twentieth Century, there was a philosophical movement stemming from Austria that aimed to do away with metaphysics. The movement has come to be called Logical Positivism or Logical Empiricism, and it is widely seen as a discredited research program in philosophy (among other fields). One of the often repeated reasons that Logical Empiricism is untenable is that the criterion the positivists employed to demarcate the meaningful from the meaningless, when applied to itself, is meaningless, and therefore it refutes itself. In this post, I aim to show that the positivists’ criterion does not result in self-refutation.
Doing away with metaphysics is a rather ambiguous aim. One can take it to mean that we ought to rid universities of metaphysicians, encourage people to cease writing and publishing books and papers on the topic, and adjust our natural language such that it does not commit us to metaphysical claims. Another method of doing away with metaphysics is by discrediting it as an area of study. Logical Positivists saw the former interpretation of their aim as an eventual outgrowth of the latter interpretation. The positivists generally took their immediate goal to be discrediting metaphysics as a field of study, and probably hoped that the latter goal of removing metaphysics from the academy would follow.
Discrediting metaphysics can be a difficult task. The positivists’ strategy was to target the language used in expressing metaphysical theses. If the language that metaphysicians employed was only apparently meaningful, but underneath the surface it was cognitively meaningless, then the language of metaphysics would consist of meaningless utterances. Cognitive meaning consists of a statement being truth-apt, or having truth conditions. If a statement isn’t truth-apt, then it is cognitively meaningless, but it can serve other linguistic functions besides assertion (e.g. ordering somebody to do something isn’t truth-apt, but it has a linguistic function).
If metaphysics is a discourse that purports to be in the business of assertion, yet it consists entirely of cognitively meaningless statements, then it is a failure as a field of study. But how did the positivists aim to demonstrate that metaphysics is a cognitively meaningless enterprise? The answer is by providing a criterion to demarcate cognitively meaningful statements from cognitively meaningless statements.
The positivists were enamored with Hume’s fork, which is the distinction between relations of ideas and matters of fact, or, in Kant’s terminology, the analytic and the synthetic. The distinction was applied to all cognitively meaningful statements. So, for any cognitively meaningful statement, it is necessarily the case that it is either analytic or synthetic (but not both). The positivists took the criterion of analyticity to be a statement’s negation entailing a contradiction. Anything whose negation does not entail a contradiction would be synthetic. Analytic statements, for the positivists, were not about extra-linguistic reality, but instead were about concepts and definitions (and maybe rules). Any claim about extra-linguistic reality was synthetic, and any synthetic claim was about extra-linguistic reality.
Synthetic statements were taken to be cognitively meaningful just if they could be empirically confirmed. The only other cognitively meaningful statements for the positivists were analytic statements and contradictions. This is an informal statement of the verificationist criterion for meaningfulness. Verificationism was the way that the positivists discredited metaphysics as a cognitively meaningless discipline. If metaphysics consisted of synthetic statements that could not be empirically confirmed (e.g. the nature of possible worlds), then metaphysics consisted of cognitively meaningless statements. In short, the positivists took a non-cognitivist interpretation of the language used in metaphysics.
Conventional wisdom says that verificationism, when applied to itself, results in self-refutation, which means that the positivists’ project is an utter failure. But why does it result in self-refutation? One reason is that it is either analytic or synthetic, but it doesn’t appear to be analytic, so it must be synthetic. But if the verificationist criterion is synthetic, then it must be empirically confirmable. Unfortunately, verificationism is not empirically confirmable, so it is cognitively meaningless. Verificationism, then, is in the same boat with metaphysics.
Fortunately for the positivists, the argument above fails. First off, there are ways to interpret verificationism such that it is subject to empirical confirmation. Verificationism could express a thesis that aims to capture or explicate the ordinary concept of meaning (Surovell 2013). If it aims to capture the ordinary concept of meaning, then it could be confirmed by studying how users of the concept MEANING could employ it in discourse. If such concept users employ the concept in the way the verificationist criterion says it does, then it is confirmed. So, given that understanding of verificationism, it is cognitively meaningful. If verificationism aims to explicate the ordinary concept of meaning, then it would be allowed more leeway when it deviates from standard usage of ordinary concept in light of its advantages within a comprehensive theory (Surovell 2013). Verificationism construed as an explication of the ordinary concept of meaning, then, would be subject to empirical confirmation if the overall theory it contributes to is confirmed.
Secondly, if one takes the position traditionally attributed to Carnap, then one can say that the verificationist criterion is not internal to a language, but external. It is a recommendation to use language in a particular way that admits of only empirically confirmable, analytic, and contradictory statements. Recommendations are not truth-apt, yet they serve important linguistic functions. So, verificationism may be construed non-cognitively, as a recommendation motivated by pragmatic reasons. There’s nothing self-refuting about that.
Lastly, one could take verificationism to be internal to a language, in Carnap’s sense, and analytic. However, the criterion would not aim to capture the ordinary notion of meaning, but instead it would be a replacement of that notion. Carnap appears to endorse this way of construing verificationism in the following passage,
“It would be advisable to avoid the terms ‘meaningful’ and ‘meaningless’ in this and in similar discussions . . . and to replace them with an expression of the form “a . . . sentence of L”; expressions of this form will then refer to a specified language and will contain at the place ‘. . .’ an adjective which indicates the methodological character of the sentence, e.g. whether or not that sentence (and its negation) is verifiable or completely or incompletely confirmable or completely or incompletely testable and the like, according to what is intended by ‘meaningful’” (Carnap 1936).
Rather than documenting the way ordinary users of language deploy the concept MEANING, Carnap appears to be proposing a replacement for the ordinary concept of meaning. The statement of verificationism is internal to the language in which expressions of meaning are replaced with “a . . . sentence of L” where ‘. . .’ is an adjective that indicates whether or not the sentence is verifiable, and thus is analytic in that language. The motivation for adopting verificationism thus construed would then be dependent on the theoretical and pragmatic advantages of using that language.
So, verificationism can be construed as synthetic, analytic, or cognitively meaningless. It could be considered a recommendation to use language in a certain way, and that recommendation is then motivated by pragmatic reasons (or other reasons), which makes it cognitively meaningless but linguistically useful, which does not result in self-refutation. Or, it could be considered a conventional definition aimed to capture or explicate the ordinary concept of meaning. It would then be verifiable because it could be confirmed by an empirical investigation into the way people use the ordinary notion of meaning, or by its overall theoretical merits. Lastly, it could be internal to a language, and thus analytic, but not an attempt at capturing the ordinary notion of meaning. Instead, it would be a replacement that served a particular function within a particular language that is itself chosen for pragmatic (non-cognitive) reasons. In any of these construals, verificationism is not self-refuting.