Philo
Sophos
·org

philosophy is for everyone
and not just philosophers

philosophers should know lots
of things besides philosophy



PhiloSophos knowledge base

Philosophical Connections

Pathways to Philosophy programs

University of London BA

Pathways web sites

Philosophy lovers gallery

GVKlempner: complete videos

PhiloSophos home

Pathways to Philosophy

Philosophical Connections

Compiled by Anthony Harrison-Barbet


CARNAP

(1891 — 1970)

 

EMPIRICISM/ LOGICAL POSITIVISM

Rudolf Carnap was born at Ronsdorf (Nordrhein-Westfalen), Germany. After his schooling at the Barmen Gymnasium he studied physics, mathematics, and philosophy at the Universities of Jena (where one of his teachers was Frege) and Freiburg. He gained his doctorate in 1921 and started his academic career as a Privatdozent at Vienna in 1926, soon becoming a leading member of the 'Vienna Circle'. In 1930 he founded the journal Erkenntnis. He was professor of natural philosophy at the German University in Prague until 1935, when he emigrated to America because of the rise of Nazism. He was appointed professor of philosophy at Chicago, and while there edited (with Neurath and Morris) the International Encyclopedia of Universal Science. In 1952 he spent two years at the Princeton Institute of Advanced Study before moving to the University of California, Los Angeles.

 

PHILOSOPHY OF LOGIC, MATHEMATICS, AND LANGUAGE

[1] [Logical Construction of the World.] Carnap's methodology and epistemology are grounded in his philosophy of language and logic; and a key concept is that of meaning. He accepted the principle of verification according to which statements are meaningful only if they are in principle verifiable; and the meaning is given by the conditions of its verification. For Carnap these conditions involve direct or indirect empirical reference through 'basic' experiences expressed in basic or 'protocol' sentences. If a statement is logically equivalent to a protocol statement, that is, can be inferred from it, the two statements are said to have the same meaning [a]. For example, 'The body S is now seeing red' (the 'physical object' sentence) is equivalent to the protocol sentence 'red, now'. Initially he adopted a somewhat 'neutral' although sceptical attitude towards metaphysical statements. But he soon regarded them in a negative light and accepted that all statements which are neither linguistic nor empirical are meaningless [Pseudo Problems in Philosophy]. The terms out of which they are constructed either lack empirical meaning or are put together in violation of syntactical rules. [See also 'The Elimination of Metaphysics'.]

These distinctions are developed further in later work [Logical Syntax of Language; see also 'The Elimination of Metaphysics']. Thus we have (i) syntactical sentences, which describe a language; (ii) object sentences, which describe physical objects; (iii) pseudo-object sentences, which look like sentences but which can be shown by analysis to be syntactical. (Statements about meaning are included in the last type.) Sentences of this third kind, which are about words rather than things, are said to be in the 'material mode'; and their real syntactical function can be shown by translating them into the 'formal mode'. For example, 'Five is not a thing but a number' can be translated as ' "Five" is not a thing-word but a number word'. It is a matter of convenience and simplicity which mode one uses. But Carnap stressed that insensitivity to use can lead to pseudo-problems, especially in relation to metaphysical and ethical statements, which cannot be reformulated as syntactical sentences, which tell us nothing about the world; they just express or arouse feelings [b]. After eliminating such pseudo-statements, philosophy thus becomes a "logic of science" — a branch of logic (or form of language), whose function is to describe the language of empirical science or to recommend changes in it [c].

[2] Central to Carnap's theory of language at this stage is his account of what he saw as the general structure or formal 'metalanguage' for characterizing the syntax of any given natural language. He allows, again as a matter of 'convenience', that a plurality of 'artificial' languages can be constructed. Every individual, he says, is free to establish his own form of language according to his requirements. This constitutes his 'principle of tolerance'. However, he says that because philosophical statements are relative to context words may, for example, may be both synonymous/ tautologous and non-synonymous/ non-tautologous, depending on the languages they belong to. Carnap therefore lays it down that meta-statements describing a language must belong to that language [a]. In this connection he distinguished 'foundation' rules, which lay down what symbols and sentences are to be allowed in the language; 'transformation' rules, which are either 'physical' (P-) or 'logico-mathematical' (L-); and two kinds of concepts — derivation terms and consequence terms. He then attempted to group all sentences into classes by reference to these rules and concepts. This can be seen in his discussion of the foundations of mathematics. Carnap broadly accepted logicism (the theory that mathematics can be reduced to logic) but sought also to incorporate elements of intuitionism (the view that classical mathematics consists of constructions which are mental, not 'Platonic') and formalism (which seeks to formalize classical and 'Platonic' mathematics from a constructivist standpoint without consideration of its intended 'meaning') [Logical Syntax]. Following the formalists, he made use of his distinction between language and metalanguage to develop two alternative 'model' languages representing respectively constructivism and classical mathematics [b].

[3] Difficulties with the principle of verification led Carnap to modify his views on meaning. What, for example, is the status of the principle itself? It is neither nonsensical nor a tautology. It cannot itself be construed as open to either verification or refutation. So he now came to think of it as a 'recommendation' for the construction of the supposedly 'ideal' language of science from which all 'metaphysical' statements have been eliminated. And correspondingly he understood meaning no longer in terms of direct or indirect verification but with reference to empirical consequences which admit of confirmation [a]. [See Probability, sec. 5.]

While in his earlier writings Carnap's aim had been to develop a theory of syntactic structures in linguistic expressions so as to exhibit more clearly how philosophical problems arise from the 'material mode' of speech, by the early 1940s [Introduction to Semantics; Meaning and Necessity], influenced by Tarski's theory of truth, he was claiming that the task of philosophy is semantic analysis, and that such concepts as 'analytic', 'synthetic', 'implication', and so on are better dealt with from a semantical rather than a syntactical standpoint. In this connection he therefore set out to formalize not only the semantic concepts of the propositional calculus ('true', 'false', and so on) but also modal concepts [b] ('necessary', 'possible'), attempting to construct a logic which would accommodate them within the semantic framework.

Carnap then modified his account of meaning still further [Meaning and Necessity and 'Empiricism, Semantics and Ontology']. Rejecting the view that linguistic expressions name entities (concrete or abstract), he supposed that they 'designated' both intensional entities (such as individual concepts, properties, or propositions — which he had earlier considered to be designated by sentences and identified with 'states of affairs') and extensional entities (such as individuals, classes, and truth-values — corresponding to the intensions). He regarded intensions as real not mental constructs, yet said they are not 'things'. Within this framework he also sought to accommodate the sense-reference distinction [c]: different descriptions of an object are said to designate different 'individual concepts' having the same extension. To deal with objections to his general account of meaning he distinguished between internal and external questions of existence. The former relate to issues that arise within the context of a particular conceptual scheme with its own appropriate empirical or logical criteria for determining what exists — in everyday matters, physics, functional contexts, and so on. However, external questions concern the nature or ontological status of the conceptual schemes themselves. It is in connection with external questions that philosophical problems arise about the existence of entities designated by linguistic expressions. But, appealing again to his principle of tolerance, he said we are free to use any language we find useful or convenient — regardless of the entities whose existence it seems to commit us to [d].

 

METHODOLOGY/ EPISTEMOLOGY

[4] Carnap set out [Logical Construction of the World] to construct a constitution system which would accommodate all knowable objects. He called this 'methodological solipsism'. He started from the concept of reducibility. By means of 'constitutional definitions' or rules concepts can be reduced to sets of 'basic' concepts [a] — provided the first concepts can be transformed into sentences containing the sets. The constitution system consists of all the definitions and theorems arranged hierarchically.

Starting with 'private-psychical' (eigenpsychische) experiences Carnap selected as basic the relation of 'remembered symmetry' and sought to construct successively the classes of quality, sense and sensation in the sensory field. Thus 'red' is definable in the class of similars which have a location in a five dimensional system. His intention was that such a constitutional system should in due course take in more complex objects — physical objects, other minds, and cultural objects. However, he allowed that it might be possible to construct a system on a wider 'physicalist' basis and which would make use of the language of' 'realism' (provided we do not accord private experiences' epistemological primacy').

This approach presented Carnap with two difficulties: (1) how private experiences can be he foundation for the public inter-verfiable sentences of science; (2) whether a 'phenomenalistic' language or a 'physicalist' language is to be preferred for the articulation of the basic 'protocol' or report sentences of science. To deal with these he moved some way from his generally positivist position and accepted that all of science (including the social or human sciences) could now be built on a physicalist basis utilizing preferably an intersubjective physicalist language into which all empirical statements (the phenomenalist language), equivalent to protocol sentences, could be expressed ['The Unity of Science'] [b].

Modifications to the concept of reducibility, particularly in relation to 'disposition' terms, then led Carnap to distinguish between definitions of scientific concepts in terms of protocol statements (whichever 'language' they might be expressed in) and reduction. Similarly he came to accept that scientific hypotheses could not be verified, only confirmed or tested. His final view therefore was that scientific concepts should be reducible to rather than definable in terms of observables; and that scientific sentences should be understood as admitting of confirmation by reference to observations instead of being translated into sentences about observables [c].

[5] Probability. In his account of probability Carnap attempted to reconcile different positions — as he had done in his treatment of mathematics. He distinguished between probability1 (confirmation probability) and probability2 (relative frequency probability) and says that they should be clearly separated ['Two Concepts of Probability'; Logical Foundations of Probability] [a]. The first is a (non-deductive) logical relation, while the second belongs to the province of statistics, involving relations between classes of events. Often what appears to be relative frequency may in fact be a statement about a logical relation between evidence and conclusion. Now, the methods for assessing the degree of 'confirmability' belong to the foundations of inductive logic. And Carnap interpreted inductive logic as parallel to deductive logic in that the methods of testing for the degree of confirmation of a hypothesis on the basis of some evidence are comparable to the rules for deriving a conclusion from premisses or theorems from axioms. He suggested there is a common feature of rationality [b] in so far as both procedures involve a recognition of meanings of sentences and of, respectively, logical implication and the definition of 'degree of confirmation'.

 

CRITICAL SUMMARY

Carnap is perhaps both the most representative and the least dogmatic of the logical positivists. His philosophy is complex and wide-ranging, his arguments and their implications being worked out in great detail. In so far as his thought was constantly undergoing development and modification in the light of criticism a summary is best presented in general rather than specific terms. He offers an 'extensional' constructionist system grounded in experience. In the early stages of his thought he was concerned largely with syntax and the primacy of phenomenal language but subsequently came to emphasize a physicalist language and semantics. He is important for the distinctions he made between language and metalanguage, the 'formal' and the 'material' mode, and 'internal' and 'external' questions, for his view that philosophical problems can be resolved by means of a logical analysis of language, and for his contributions to probability theory. Some of the main criticisms which can be made of his philosophy are the following.

(1) At the level of 'protocol' sentences, which are about supposedly incorrigible observation reports, he would seem to be committed to some form of correspondence theory of truth. Some critics, however, have questioned whether Tarski's semantic theory of truth is a genuine correspondence theory in that his formal definition of truth seems to have no application to natural languages [compare also Davidson]. It might therefore be argued that there is a tension here in Carnap's philosophy. It seems also that there is a tacit appeal to coherence as a test of truth for sentences other than protocol sentences (a sentence being true by virtue of its relations to other sentences). Furthermore, his principle of tolerance and acceptance of different conceptual schemes would appear to introduce both pragmatic and relativist features. Against all this it could be said that it is a strength rather than a weakness that these different positions are accommodated within his philosophy of language and his epistemology.

(2) The central feature of the positivist/ empiricist programme — the reduction or assimilation of one mode of discourse to the scientific has been criticized — particularly by 'ordinary language' analytic philosophers, who claim that this programme cannot succeed.

(3) These philosophers have also questioned the tenability of Carnap's view that linguistic expressions 'designate' entities (extensional and intensional) the existence of which seems to be a matter of which conceptual scheme is adopted. The supposition in his later writings that entities have a 'Platonic' existence must also contend with 'nominalist' objections.

 

READING

Carnap: [of many writings] Die Logische Aufbau der Welt (1928) (The Logical Construction of the World, trans. R. A. George); Scheinprobleme in der Philosophie: Das Fremdpsychische und das Realismusstreit (1928) (Pseudo Problems in Philosophy: Other Minds and the Realism Debate, trans. R. A. George); Logische Syntax der Sprache (1934) (Logical Syntax of Language, trans. A. Smeaton); Introduction to Semantics (1942); Meaning and Necessity: A Study in Semantics and Modal Logic (1947; 2nd edn 1956); Logical Foundations of Probability (1950). See also these essays: 'berwindung der Metaphysik durch die logische Analyse der Sprache' (1931/2) ('The Elimination of Metaphysics through Logical Analysis of Language', trans. A. Pap, in A. J. Ayer, Logical Positivism); 'Die Physikalische Sprache als Universalsprache der Wissenschaft' (1932) ('The Unity of Science', trans. M. Black); 'The Two Concepts of Probability' (1945), reprinted in H. Feigl and W. S. Sellars, Readings in Philosophical Analysis); 'Empiricism, Semantics, and Ontology' (1950), reprinted in Meaning and Necessity, 2nd edn.

Studies

N. Goodman, The Structure of Appearance [concentrates on Der logische Aufbau].

A. W. Richardson, Carnap's Construction of the World: the Aufbau and the Emergence of Empiricism.

Collections of essays

P. A. Schilpp (ed.), The Philosophy of Rudolf Carnap.

J. Hintikka (ed.), Rudolf Carnap, Logical Empiricist: Materials and Perspectives.

B. H. Kazemeier & D. Vuysje (eds), Logic and Language.

 

 

CONNECTIONS

Carnap

 

Note: Attention should be paid to Otto Neurath (1882-1945), a co-worker of Carnap in the 'Vienna Circle' and another proponent of physicalism. Note also the general influence on Rorty of Carnap's approach to philosophical problems.

 

[1a; cf. 3a c] Statements meaningful if verifiable (meaning given by conditions of verification); synonymy and logical equivalence — basic 'protocol' statements

   Russell

Schlick

   Wittgenstein

Popper

Hempel

Quine

Ayer

Dummett

[1j]

[2d]

[3a]

[1e]

[1a]

[3a]

[1a d 2b 1b]

[1f]

 

[1b] Metaphysical statements meaningless (neither empirical nor 'linguistic'), or syntactically erroneous; (later) cannot be reformulated as syntactical sentences

Schlick

   Wittgenstein

Popper

Ayer

[2a b e]

[3a]

[1d]

[1a 3a]

 

[1c] Philosophy as 'logic of science', i.e., language to describe, improve language of empirical science

   Frege

Schlick

Wittgenstein

Ayer

Dummett

[2b]

[1d]

[3a]

[1e]

[1a]

 

[2a; cf. 2b 3d 4b] Metalanguage to describe natural language (and belongs to the language); tautology/ synonymy relaative to language; plurality of languages constructible; 'principle of tolerance'

   Russell

Schlick

   Wittgenstein

[1h 2d]

[1d 2b]

[1b 1d 2b 3c]

 

[2b; cf. 2a] Mathematics reducible to logic; constructivism, and classical as alternative 'languages' (conventionalism)

   Berkeley

   Frege

   Russell

   Wittgenstein

Ayer

[1d]

[1a c]

[1a]

[1e 2f]

[1a]

 

[3a; cf. 1a 3c 4c 5a] [Later] meaning in terms of empirical consequences; confirmation rather than verification

   Russell

Schlick

Popper

Hempel

[1i j]

[1b 2d e]

[1e]

[2b]

 

[3b] [Later] philosophy as semantic analysis; analytic-synthetic distinction and modal logic semantically considered

Quine

[1b]

 

[3c; cf. 3a] [Later] meaning: linguistic expressions do not denote concrete or abstract entities but 'designate' intensional (real but not 'things') and extensional entities; propositions not now 'states of affairs'; sense-reference distinction in terms of 'individual concepts' with same extension

   Frege

   Russell

   Wittgenstein

Ryle

Hempel

Quine

Ayer

[2c e]

[1i]

[1a c 2a]

[2a]

[1c]

[1a 2a 3a]

[1c]

 

[3d; cf. 2a] 'Internal' and 'external' questions — status of 'existents' within conceptual schemes and of the schemes themselves'; principle of tolerance (convenience)

Popper

Quine

   Davidson

[2a]

[2a]

[1e]

 

[4a; cf. 1a] 'Methodological solipsism'; 'constitution system' accommodates knowable objects; reducibility to basic concepts through rules

   Russell

   Wittgenstein

Quine

[1a 2b 3b]

[1a]

[3a]

 

[4b; cf. 2a b] [Later] all sciences (natural and 'human') built on physicalist basis; empirical (cf. protocol) statements (phenomenalist language) translatable into intersubjective physicalist language

   Russell

Schlick

Popper

Hempel

Quine

Merleau-Ponty

Ayer

[2b 3b]

[1c 2f]

[3b]

[1c 2a c]

[3a]

[3a]

[2a c 3c]

 

[4c; cf. 3c 5a] Scientific hypotheses confirmable rather than verifiable; scientific concepts, observables, and translatability

Popper

Quine

Ayer

[1a 2a b]

[3a]

[4d]

 

[5a] Confirmation probability and relative frequency probability distinct; degree of confirmability foundation of inductive logic

Popper

Hempel

Ayer

[1c]

[2b]

[4e]

 

[5b] Inductive and deductive logic parallel; testing methods comparable to deductive rules — common rationality

   Mill

Popper

Hempel

Ayer

   Strawson

[1g]

[1a]

[2b]

[4c-e]

[1g]