The Institutionalized Aim of Science

          The preceding sections have discussed the archetypal twentieth-century philosophies of science: Romanticism, Positivism, Pragmatism, and Psychologism.  And they have also discussed the basic perspectives of language: syntax, semantics, ontology, and pragmatics.  Finally consider next the four topics in philosophy of science in the light of these previous discussions beginning with the institutional aim of science.
          Issues about the aim of science are the most fundamental, because they profoundly affect all the other topics.  And as it happens the literature of philosophy of science offers a variety of proposals for the aim of science.  The Positivists had proposed that science should achieve firm foundations either by relying on observation language exclusively or by limiting theoretical terms to those that are related by logical reduction to an observation language serving as a reduction base.  Neurath, a proponent of the unity of science agenda, proposed that all sciences including the social sciences aim at logical reduction to physics, which in turn is to be reduced to observation. On the other hand Romantics in the social sciences maintain that the sciences of nature differ fundamentally from the sciences of culture, which are the social sciences.  They propose that science aims at vicarious imputation of subjectively based interpretative “understanding”, so that an explanation “makes sense” to the social scientist due to his personal experiences as a participant in shared human nature and, when possible, participation in the same culture as the social agents he is studying.  Some of them advocate the philosophy of Weber, in which this understanding called “verstehen” is not only a source for the requisite mentalistic ontology, but is also a basis for validation.  Most fundamentally Romantics who do not altogether reject the aim of prediction and control in cultural sciences, subordinate it to interpretative understanding.
          Most of the more recent proposals in academic philosophy of science arise from reflection on episodes in the history of the natural sciences.  Popper, reflecting on the development of relativity theory by Einstein, proposed that the aim of science is to produce tested and nonfalsified theories having greater information content than their predecessors.  Kuhn, reflecting on the development of the much earlier Copernican heliocentric theory, proposed that small incremental changes extending the prevailing theory defines the institutionalized aim of science, which he called “normal” science, and that scientists do not consciously aim to produce revolutionary new theories.  Feyerabend, reflecting on the development of the quantum theory, proposed that each scientist has his own aim, and that anything institutional is an impediment to science.  His philosophy of science is an early variation upon the then-emerging Pragmatist ideas, but it is also a quite idiosyncratic version. Thagard, reflecting on the wave theory of sound and on other more recent developments in natural science, proposed that scientists choose theories that maximize what he calls “explanatory coherence”, which he defined in terms of empirical adequacy, breadth of explanation, simplicity of explanation, and analogy with established explanations.  He developed his computerized cognitive system ECHO, to simulate the realization of this aim in various episodes of theory choice in the history of science. 
         The contemporary Pragmatist philosophy is now the ascendant view in academic philosophy.  It evolved from an examination of the development of quantum theory in physics in the 1920’s and from a consequent critique of Positivism.  However, the mature articulation of the contemporary Pragmatism did not come to fruition until the early 1970’s.  Today Pragmatists view modern empirical science as a cultural institution having its distinctive system of views and values.  The institutionally regulated activities of research scientists may be described succinctly in a statement of the aim of science, which the contemporary research scientist seeking to maximize his success may employ as what some social scientists call a rationality postulate. The Pragmatist rationality postulate for the practice of research in the empirical sciences is the following statement of the aim of science:

Scientists aim to construct explanations by developing theories that satisfy the most critically empirical tests that can be applied at the current time.  Such satisfactory theories may be called scientific laws.

This statement is explained by examining the second, third, and fourth topics in philosophy of science as three sequential steps.  It can be rephrased to describe the successful achievements in the history of science, so as not to impute motives to scientists whose personal objectives and psychological experiences often cannot correctly be described in a statement of the conscious aim of science.  The statement rephrased in terms of successful outcomes instead of a conscious aim reads as follows:

Science achieves explanations by developing theories that satisfy the most critically empirical tests that can be applied at the current time. Such satisfactory theories may be called scientific laws.

Institutional Change

         Change of institutions is different from change within institutions.  In the history of science successful researchers in basic science have routinely failed to understand the reasons for their success, and have often formulated or accepted erroneous philosophies of science to explain their successes.  One of the most historically notorious such misunderstandings is Newton’s “Hypotheses non fingo”, his denial that his monumental theory of gravitation is a hypothesis.  In due course such false practices and beliefs become suspect, as successful developments are achieved in spite of the erroneous proscriptions and prescriptions.  As Feyerabend noted in his Against Method, successful scientists have often broken prevailing methodological rules.  The successful and institutionalized practices of scientific research had firstly to evolve through trial and error before they could be examined, analyzed, and formulated into new philosophies of science. The rationality postulate is therefore a postulate in the sense of a hypothesis, and what is rational today will likely be seen tomorrow as superstition, as both science and philosophy of science continue to evolve.  Not surprisingly there exists what may be called a cultural lag between the evolution of science and the development of philosophy of science, since the latter depends on the former.  For example over thirty years passed between the development of the modern quantum theory and the consequent emergence of the contemporary Pragmatist philosophy of science. The evolution in science that involves a revision of the rationality postulate amounts to an institutional change.  Such changes do not occur rapidly or easily, and are usually intergenerational due to the magnitude of the adjustment.
         Not only is there a cultural lag between science and philosophy, there are also cultural lags among the several sciences.  Philosophers of science have preferred to examine physics and astronomy, because these have been the most advanced of the sciences since the historic Scientific Revolution, which started with Copernicus.  Many other sciences have tended to lag behind physics and astronomy with the social and behavioral sciences farther behind than the natural sciences other than physics and astronomy.  The result has been the survival of philosophical superstitions in the lagging sciences, especially to the extent that they have looked to their own less successful histories to formulate their own philosophies of science.  For example sociologists and many neoclassical economists continue to use a Romanticist philosophy of science, and believe that cultural sciences or sciences of “human action” are fundamentally different from the natural sciences.  In addition, the behaviorist school of psychology continues to use the Positivist philosophy of science.  In the contemporary perspective these sciences are institutionally retarded, because they impose prior ontological commitments – either mentalistic or nonmentalistic - as criteria for scientific criticism.
         Institutional change in science must be distinguished from change within the prevailing institutional matrix of the aim of science and the criteria for scientific criticism. Philosophy of science is principally concerned with the latter.  It has less to say about the former except retrospectively, because institutional change is unique and distinctively historical.  Its occurrence can be recognized retrospectively, because it is seen to involve not only a change in formerly accepted explanations in science, but also a change in the prevailing concept of the nature of science itself.  Contrary to philosophers such as Kuhn, the existing institutional matrix is not identified with the prevailing scientific explanations.  There have been revolutionary developments in science such as Darwin’s theory of evolution that had no effect on the institution of basic science, however great the impact Darwin’s theory had on the science of biology and on the macrosociety.  In fact it is the enduring stability of the institution of science through even dramatic revolutionary changes that makes philosophy of science possible and useful to the practitioner of basic research science.

Scientific Discovery

          Recall the distinctively Pragmatist meaning of the term “theory” as universally quantified statements proposed for testing.  The topic of scientific discovery is the problem of creating new theories that will pass empirical testing with nonfalsifying outcomes.  There have been other ideas about discovery depending on the meaning of “theory.”  Positivist philosophers’ discussion of this topic consisted of induction, which yields empirical generalizations, and of the human creative processes, which yield theories.  But they could offer no explanation as to how scientists create theories.  For Positivists the term “theory” refers to sentences containing “theoretical terms”, which describe unobserved entities.  Such entities can be microphysical particles such as electrons or mental states such as ideas.  For Romantic social scientists and philosophers the creative process consists of the imputation of vicariously based ideas and motives that “make sense” to the social scientist because he can recognize them in his personal experience.  Thus the social sciences are cultural sciences in which the term “theory” refers to language describing the mental states experienced by the subjects of their social theories. On the contemporary Pragmatist view there is no separate class of vocabulary called “theoretical terms”, as the Positivists thought, nor do mental experiences warrant uniquely labeling discourse about it “theory”, as the Romantics thought.  For the contemporary Pragmatist philosophers “theory” is defined pragmatically instead of semantically; it is any universally quantified discourse proposed for empirical testing.  Thus the problem of scientific discovery is essentially that of analyzing and proceduralizing the creation of such statements that are empirically testable and hopefully when tested are not falsified. 
         As mentioned above both theory development and theory testing change the state description of the language in the science, and thus offer a dynamic diachronic view.  Theory creation introduces new language into the current state description, while falsification eliminates language from the current state description.  The most significant work addressing the problem of scientific discovery has been the relatively recent development of computerized discovery systems. These systems, also called “artificial-intelligence” systems, describe the transition from an inputted state description to an outputted one generated by the computer system and representing a later language state.  To be useful every discovery system must contain procedures both for theory creation and for theory selection.  Different computer systems created by different developers implement different strategies in their system designs for the discovering.  If the discovery system is a generative grammar, then only the descriptive vocabulary from the initial state description is inputted to the system.  But whatever the system design, the input information is from an initial state description, and the output information is the terminal state description.  There are issues in the philosophy of science literature as to just what the state descriptions are describing.  On the cognitive psychology agenda, the state descriptions represent in individual’s psychological state consisting of mental representations.  On the linguistic analysis agenda, the state descriptions represent the shared semantics of a language-using community constituting a scientific profession.  On either interpretation, however, the input state description represents the knowledge available for future discovery, and the output state description is the one or several new theories that constitute the discovered knowledge.
         The sources of language for the input state description is crucial for a discovery system.  In his Introduction to Metascience (1976) Hickey distinguished three types of theory development that are relevant to input language: (1) theory extension, (2) theory elaboration, and (3) theory revision.  Firstly theory extension is based on a currently nonfalsified theory that is used to address the scientific problem under investigation.  The extension could be a simple addition of statements to make a general theory more specific for a new problem.  This process involves minimal change.
         Secondly theory elaboration is the correction of a currently falsified theory by the addition of new factors or variables in a manner that changes the theory's predictions while preserving the theory’s universal quantification so it is not merely ad hoc.  The input language consists of factors or variables that represent anything that seems plausible for solving the problem, and the amount of vocabulary inputted to a mechanized discovery system could be large.  This theory-development strategy amounts to a fishing expedition in search for a correcting factor or variable. 
         Thirdly theory revision is essentially a reorganization of the constituent information in existing theories.  The source of input for theory revision consists of the descriptive vocabulary from all the currently nonfalsified theories addressing the problem at hand.  The nonfalsified theories need not have been tested empirically.  Since the problem is unsolved, it does not have any theory that is tested and not falsified.  The descriptive vocabulary from recently falsified theories may also be included as inputs to make an accumulative state description.  Rejected theories have scrap value.    The size of the input state description is relatively small. Yet it must be large enough to supply sufficient information for the development of a new theory.  The new theory is typically very different from previous theories. This output is most likely to be called “revolutionary.”  Hickey’s METAMODEL system has been used for both theory elaboration and theory revision, often combined in the same input.
         The revision can also be the patterning of a proposed solution to the new problem by analogy with an existing explanation.  Thagard’s reconstruction of the development of the theory of sound waves on analogy with water waves by means of his PI system might be taken as an example of mechanized theory revision.  This source of input for analogy, however, is potentially very large, and this strategy has not been used in any mechanized system for developing a contribution to the current state of any science, although there are many examples of the use of analogy in the history of science.
         To date discovery systems that have actually produced new theories for a scientific profession have had certain characteristics. Firstly researchers working in their own specialized scientific field of application have developed the effective discovery systems, while neither academic philosophers nor cognitive psychologists have such a track record. Cognitive psychologists have been content to apply their discovery systems to the replication of past episodes in the history of science, rather than apply their systems to the current state of a science and actually produce new theories.  Their efforts to date have been like a stage play in perpetual rehearsal with no performance.  Secondly the discovery procedures used in the systems are typically described as merely the mechanized automation of theory-developmental practices already used in the scientific field of application.  Thirdly the input descriptions contain numerical data, and the mechanized discovery procedures applied to the input data incorporate statistical-analysis procedures.  Fourthly and finally the scientific fields of application have been the social sciences.  Statistical inference procedures are commonly used in the social sciences to discover relations among data, thus making these sciences obvious opportunities for the first useful discovery systems. 

Scientific Criticism

         The philosophical discourse on scientific criticism has little to say about the specifics of experimental design.  Instead it pertains to the criteria for the acceptance or rejection of theories. The only criterion acknowledged by the contemporary Pragmatists is the empirical test.  Whenever in the history of science there has been a conflict between the empirical criterion and any nonempirical criteria for the assessment of new theories, eventually it was always the empirical criterion that governed theory selection.  Contemporary Pragmatists accept scientific realism and ontological relativity, and therefore reject all prior ontological criteria and subordinate ontological commitment to empirical criticism.

The Logic of Testing

         The universally quantified statements of the theory in an empirical test can be cast into a conditional proposition in the form “If A, then C.”  The antecedent clause “A” represents the set of universally quantified statements describing the antecedent conditions, those of the test-design for the test.  When the test is executed the logical quantification of “A” is changed to particular quantification to describe the individual test situation, and it is regarded as true, if the test is executed in compliance with its test design.  The empirical test is conclusive only if it is executed in accordance with its test design. 
         The consequent clause “C” represents the set of universally quantified statements describing the predicted outcome of the execution of a test.  Its logical quantification is changed to particular quantification to describe the predicted outcome of the individual test.  Another statement, “O”, which also has particular quantification, describes the observed outcome from execution of the test in the same vocabulary that is used in the prediction statement “C.”  The logic of the test is the nontruth-functional modus tollens argument form, according to which the conditional hypothetical statement expressing the theory is falsified if the statements “C” and “O” are not accepted as saying the same thing, i.e. if the prediction is wrong. 
         The nontruth-functional conditional logic implements Popper’s falsificationist philosophy of scientific criticism.  The conditional statement expressing the tested theory asserts not merely a conjunction, but a dependency between the phenomena described by the antecedent and consequent components.  This claimed dependency cannot be conclusively established or verified on the basis of the truth-values of the component statements except in the case of falsification. The truth table for the truth-functional Russellian logic therefore is not the logic of empirical testing in science. When the antecedent clause is false, the test is invalid due to a failure to comply with its test design.  For purposes of comparison truth-functional and nontruth-functional truth tables appear as follows:

                                            Nontruth-Functional

                           Truth Table                                      Truth Table

                    A          B          A É B                 A          B          If A, then B.

                     T         T           T                        T          T          Not Falsified

                     T         F           F                        T          F          Falsified

                     F         T           T                        F          T          Invalid Test

                     F         F           T                        F          F          Invalid Test

Empirical Decidability and Semantics

         The decidability of empirical testing is not absolute.  Popper had recognized that the statement reporting the observed test outcome, which he called a “basic statement”, requires prior agreement among the cognizant scientists, and that it is not incorrigibly true.  Normally the semantics is such that if a test has a nonfalsifying outcome, the semantics is unchanged with the universally quantified statements of both the theory and the test design contributing to the meanings of the terms common to both kinds of statements.  But when the outcome is a falsification, there is a semantical change produced for those who accept the outcome as a falsification of the theory.  The test-design statements continue to control the semantics of the terms common to the theory and test design by contributing their parts of the meaning complex of each of the common terms.  But the parts of the meaning complex contributed by the theory statements are excluded from the semantics of those common terms, at least for those who previously believed in the tested theory but no longer do as a result of the test.
         In the event of falsification, there is also a different semantical change produced for those who do not accept the outcome as a falsification of the theory.  Such a dissenting scientist has reconsidered either the test-design statements or the report of the test outcome.  If he challenges the test outcome, then he has merely questioned whether or not the test was executed in compliance with its agreed test design, and the test may be repeated to answer his challenge to validity. 
         But if he challenges the test design itself, then he has changed his mind about the test design, and has thereby changed the semantics involved in the test in a fundamental way.  The semantical change produced for such a recalcitrant believer in the theory consists in the theory statements controlling the meanings of the terms common to the theory and test-design statements.  In that case the parts of the meaning complex contributed by the test-design statements are the parts excluded from the semantics of at least one and probably several of the terms common to the theory and test-design statements.  This amounts to attacking the test design as if it were falsified, and letting the theory define the subject of the test – a role reversal in the pragmatics of the test design and theory language, that redefines the problem under investigation.  Popper rejects such a response to a test, calling it a content-decreasing stratagem, and he admonishes the scientist to stick to his problem and refrain from criticizing everything.  But the dissenting scientists may decide that the design of the falsifying test is a misconception of the problem that the tested theory is intended to solve, and may take exception to a measurement procedure or other aspects of the test design.  Empirical tests are conclusive decision procedures only for the scientists who agree on which language is proposed theory and which language is presumed test-design.  

Pages [1] [2] [3] [4] [5] [6]
NOTE: Pages do not corresponds with the actual pages from the book