3.43 Pragmatic Definition of Theory Language
Scientific theories are universally quantified statements including mathematical expressions (a.k.a. “models”) that are proposed for empirical testing.
Unlike positivists and romantics, pragmatists define theory language pragmatically, i.e., by its function in research, instead of syntactically as an axiomatic system or semantically by some distinctive content. This functionality supplies the definition of “theory” in the contemporary pragmatist philosophy of science. It contains the traditional idea that theories are hypotheses, but the reason for their hypothetical status is not due to either the positivist observation-theory dichotomy or the romantics’ requirement of referencing subjective mental states. Theory language is hypothetical because interested scientists agree that in the event of falsification, it is the theory language that is falsified instead of the test-design language. Often theories are deemed to be more hypothetical, because their semantics is more empirically underdetermined.
Theory is a special function of language – empirical testing – rather than a special type of language.
Scientists believe that the proposed theory statements are more likely to be productively revised than the presumed test-design statements, if a falsifying test outcome shows that revision is needed.
Pragmatically after a theory is tested, it ceases to be a theory, because it is either scientific law or rejected language, except for the skeptical scientist who wants further predictive testing. Designing empirical tests can tax the ingenuity of the most brilliant scientist, and theories may have lives lasting many years due to difficult problems formulating or implementing decisive test designs. Or as in a computerized discovery system with an empirical decision procedure, theories may have lives measured in milliseconds.
After a conclusive test outcome, the tested theory is no longer a theory, because the conclusive test makes the theory either a scientific law or falsified discourse.
Romantic social scientists adamantly distinguish theory from “models”. Many alternative supplemental speculations about motives, which they call “theory”, can be appended to a model that is has been tested. But it is the model that is empirically tested statistically or predictively. Pragmatically the language that is proposed for empirical testing is theory, such that when a model is proposed for testing, the model has the status of theory.
Some time after initial testing and acceptance, a scientific law may revert to theory status to be tested again. Centuries after Newton’s law of gravitation had initially been accepted as scientific law, it was tested in 1919 in the famous Eddington eclipse test of Einstein’s alternative general relativity theory. Thus for a brief time early in the twentieth century Newton’s theory was pragmatically speaking actually a theory again.
Thus the term “theory” is ambiguous; archival and pragmatic meanings can be distinguished. In the archival sense philosophers and scientists still may speak of Newton’s “theory” of gravitation, as is often done herein. The archival meaning is what in his Patterns of Discovery Hanson calls “completed science” or “catalogue science” as opposed to “research science”. The archival sense has long-standing usage and will be in circulation for a long time to come.
But research scientists seeking to advance their science using theory in the archival sense are seeking an El Dorado. The archival sense is not the meaning that is needed to understand the research practices and historical progress of basic science. That is why philosophers today recognize the pragmatic meaning of “theory”, which is a transitional phase for a science. In the pragmatic sense Newton’s “theory” is now falsified physics in basic science and is no longer proposed for testing, although it is still used by aerospace engineers and others who can exploit its lesser realism, i.e., lesser truth.
3.44 Pragmatic Definition of Test-Design Language
Pragmatically theory is universally quantified language that is proposed for testing, and test-design language is universally quantified language that is presumed for testing.
Accepting or rejecting the hypothesis that there are red ravens presumes a prior agreement about the semantics needed to identify a bird’s species. The test-design language defines the semantics that identifies the subject of the tested theory and the procedures for executing the test design. This semantics includes but is not limited to the language for describing the design of any test apparatus, the testing methods including any measurement procedures, and the characterization of the test’s initial conditions. The semantics for the independent characterization of the observed outcome resulting from the test execution is also defined in the test design language. The universally quantified test-design statements contribute these meaning components to the semantics of the descriptive terms common to both the test design and the theory.
Both theory and test-design language are believed to be true, but for different reasons. Experimenters testing a theory presume the test-design language is true with definitional force for identifying the subject of the test and for describing the procedures for executing the test. The advocates proposing or supporting a theory believe the theory statements are true with sufficient plausibility to warrant the time, effort and cost of testing with an expected nonfalsifying outcome. For these advocates both the theory statements and the test-design statements contribute component parts to the complex semantics of the descriptive terms that the theory and test-design statements share prior to testing.
Often test-design concepts describing the subject of a theory are either not yet formulated or are too vaguely described and conceptualized to be used for effective testing. They are concepts that await future scientific and technological developments that will enable formulation of an executable and decisive empirical test. Formulating a test design capable of evaluating decisively the empirical merits of a theory often requires considerable ingenuity. Eventual formulation of specific test-design language enabling an empirical decision supplies the additional clarifying semantics that sufficiently reduces the disabling empirical underdetermination in the descriptive terms of the theory.
3.45 Pragmatic Definition of Observation Language
Observation language is test-design sentences that are given particular logical quantification for describing an individual test procedure and execution including the reporting of the test outcome.
After scientists have formulated and accepted a test design, the universally quantified language setting forth the design determines the semantics of its observation language. Particularly quantified language cannot define the semantics of descriptive terms. The observation language in a test is sentences or equations with particular logical quantification accepted as experimentally or experientially true and used for description, and it includes both the test-design sentences describing the initial conditions and procedures for an individual test execution and also the test-outcome sentences reporting the outcome of an executed test. This is a pragmatic concept of observation language, because it depends on the function of such language in the test. Contrary to positivists and earlier philosophers, pragmatists reject the thesis that there is any inherently or naturally observational semantics.
If a test outcome is not a falsification, then the universally quantified theory is regarded as a scientific law, and the theory contributes its semantics to the meaning complex associated with the descriptive terms in the universally quantified test-design sentences. Additionally the terms in the universally quantified test-design sentences contribute