Science describes and explains the world in indirect ways, i.e., its access to phe-
nomena is never direct. We will presuppose this as an essential aspect of science, without
further argument at this stage. To avoid the possibility of misunderstanding, we would like to warn the reader that the distinction between direct and indirect made here is not on the
basis of sensory experience, but on the basis of whether a description is made dependent or
independent of astructure. What precisely is the nature of the structure shall be elaborated below.
As mentioned already, when scientists study ‘something’, an object of investigation,
12
Different versions of statement views and nonstatement views are identified by Stegm¨uller 1979, The Structuralist View of Theories: A Possible Analogue of the Bourbaki Programme in Physical Science p. 4ff.
128 Chapter 5. Nature and Structure of Scientific Knowledge
they do not describe all aspects of this ‘something’; rather they select by way of abstraction certain parameters, among others, from this ‘something’. The abstraction consists in orga- nizing (structuring) the selected parameters into a system, which are ‘lifted’ from the rest of the ‘something’. Such systems, as some would like to call, are idealizations. Only under experimental conditions, and with considerable approximations, can these systems be real- ized in the actual world. When phenomena are described using these structured parameters, themeaning of the description is not independent of the constructively visualized system or
structure, and hence it is an indirect description or observation. Since scientific description
cannot be true of the world without the involvement of idealization, the semantics of scientific
knowledge demandscounter-factual interpretation. Let us call a description scientific if and
only if it is intelligible (meaningful) only through indirect (structure dependent) means. Now
since a scientific assertion accurately describes ‘something’ only under ideal conditions, as
stipulated by the structural relations, it is not true under normal circumstances. The cir-
cumstances in which a scientific assertion is entirely correct or true is called a model. This definition of a model is a micro-version of the usual definitions of a model. Differences with the usual definition, and the reasons for deviation are given below.
A physical system can be viewed as an idealized replica of the phenomena, which
can be specified solely in terms of the selected parameters.13
Examples of physical systems abound in science, and can be found in all disciplines: different kinds of instances of New- tonian particle systems, atomic system, dynamic systems such as oscillating or vibrating systems, thermodynamic systems, ideal gas systems, chemical equilibrium systems, physi- ological systems like the nervous, endocrine, circulatory systems, etc. However, there are other simpler and more general systems that scientists regularly employ at various levels of theorization, such as lever, balance, floating bodies, pendulum, etc. The role played by these simpler systems in the initial stages of the development of scientific knowledge is elaborated in greater detail below in Part-III. Most reconstructions of scientific theories have been at- tempted for relatively complex theories such as classical particle mechanics. We think that scientists started constructing, defining and using systems much earlier than the 17th cen- tury and can be traced back to the early origins of geometry. Our attempt in the present thesis will be to understand the genesis, development and the structure of simpler systems, by applying the semantic approach.
The function of models, in relation to physical systems, can be stated as follows:
Models are employed to represent the behavior of a certain kind of physical system. Models
13
are usually devised as mathematical structures whose characteristics are obtained or specified by definition. For example, a Newtonian particle ‘system’ is regarded as a model because it
is a structure that satisfies the three laws of motion and the law of universal gravitation.14
Here the term ‘system’ appears in the name of a model. This terminology is unfortunate, but since most of us are used to calling most models by the name ‘system’ we shall continue to use the same terms. Whatever the term be, our criteria for characterizing models and physical systems is precise, which will be elaborated below.
Models can be defined with various degrees of complexity. For example, a Newtonian model can be defined for a 2-body system, which can be used to represent the physical system, Earth-Moon system, or it can be defined as a general model for a n-body system that can be used for a range of simple to complex physical systems. Given certain constant values to the parameters (initial conditions) the exact behavior of the physical system can be known for certain systems deterministically, while for others probabilistically, though in both cases this knowledge holds accurately only for an ideal physical system.
Though very general models can be constructed we will attend to very simple models, and also to very simple physical systems. This makes our problem of understanding the process of model building easier. Though the models and the systems that we will choose are simple they have sufficient complexity so that whatever we could state for them could without much problem be generalized to more complicated models and systems. However, we have other reasons for choosing simpler elements for analysis. Some of the reasons can be spelled out here, while certain others will be stated in the process at appropriate places.
Earlier (page 128) we have stated that the circumstances in which a scientific asser- tion is entirely correct or true is called a model, and also mentioned that this definition is a micro-version of the usual definition. The difference can be stated by comparing it with van Fraassen’s definition, which is as follows:
A model is called a model of a theory exactly if the theory is entirely true if
considered with respect to this model alone.15
While we are talking in terms of ‘scientific assertion’, van Fraassen is talking in terms of a ‘theory’. The motivation in deviating from the usual definition is not fundamental, though highly significant for our purpose. It is, first, to make the units of semantic analysis of scien- tific knowledge simpler. Secondly and most importantly, our attempt is to use terminology that needs least mention of the expressions ‘theory’ or ‘theoretical’. The term ‘theory’ is
14
Cf. Giere 1984,Understanding Scientific Reasoning,pp. 80-81.
15
130 Chapter 5. Nature and Structure of Scientific Knowledge
as problematic as the term ‘law’, and more or less for the same reasons. Van Fraassen, has argued rather convincingly that without using the problematic term ‘laws’ we can talk about
and appraise scientific theories.16
We are proposing to move a step further, in search of neu- tral terminology, by suggesting the use of the term ‘scientific assertion’ in place of the term ‘law’ (though not always in place of the term ‘theory’). We see a possibility of describing the structure of scientific knowledge in terms of definitions, models, physical systems, scientific assertions etc., without leaving much residue, hence without requiring to talk in terms of ‘theory’. The usefulness of our proposal towards acquiring a neutral vocabulary will be made clearer as we develop our views.
The first reason for choosing a smaller unit of analysis is based on the needs of a generativist. The usual examples of scientific theories, such as Newtonian mechanics, electromagnetic theory, relativity theory etc., are ‘huge’ structures, and it is therefore difficult to comprehend them by means of reconstruction in a significant manner. Besides, looking at these theories as one unitary or holistic structure has led to problems of a serious nature, specially in understanding the relationships, such as reduction, between different theories. We have also come to know that all applications do not involve the entire structure of the theory. More often very small components of the theory are employed to deal with some cases. From the point of view of a generativist, viewing a scientific ‘theory’ as one whole, and attempting to find out how such a thing could have been discovered makes it a very complicated problem. A generativist’s strategy, we think, has to be, to start from the units and proceed to the whole—to understand how the whole can be constructed out of the units. Therefore, most examples of reconstruction of bits of scientific knowledge that we shall elaborate below will be local case studies of a given field of science. This approach, we shall argue, will be more promising than attempting to understand the discoverability of mega-structures as one piece. We now offer some clarification regarding the notion of ‘scientific assertion’.
A scientific assertion, in simple terms, is a statement relating a model—a type— with a physical system—a token. For example, in “Earth-Sun system is a two-body model (of Newton’s theory)” the system and the model take the places of the subject and the predicate positions of a statement. The term ‘Model’ can be called a predicate (more precisely a
complex predicate) because it describes aclass of physical systems, all of which can be called
the tokens of the model. Therefore, another way of defining a scientific assertion is that it is a statement where the subject is a physical system and the predicate is a model. A similar
16
In the first part of his bookLaws and Symmetries this thesis is argued in great detail. Though relevant to the matters we are discussing we shall not enter into that involved debate here.
characterization is also possible for meta-scientific statements, with a minor qualification (to be elaborated below). A scientific assertion differs from ordinary (commonsense) assertions in
a special sense that over and above the type-token relation, scientific predicates and subjects
also have a structural or morphic relation between them. This enables us to say that to have a scientific knowledge of “some thing” is to have a physical system that incorporates the phenomena which is described by an assertion employing a model. Nothing can be called scientific if the description is not structure dependent, i.e., indirect. This is our general criterion of demarcation.
Since both systems and models are constructions by the scientist, one might say,
they say nothing about thereal world. There is a partial truth in this, because while a model
is constructed by employing mathematical methods and not by empirical means, physical systems are idealizations of phenomena. However, the possibility that some of the physical systems can be really actualized in the experimental world, if not in the ‘open’ world, provides sufficient reason for our belief in the applicability of the model to describe and explain the world around. Though due to the employment of mathematical techniques scientists can obtain models that may not have any known applications, the models obtained will have certain epistemic value because of their ability to tell us under what conditions an application of such a model could be found. A number of examples in modern physics can be cited, specially in particle physics, where scientists have devised a mathematical model prior to any empirical confirmation. The predictions that are based on such models were experimentally realized much later.
The question however remains, can there be any method or logic for generating or constructing models and systems? Since models and systems are different, and since both of them are constructions, the question to be addressed by a generativist must be able to meet both the requirements, either separately or together. In fact this is a consequence of any generativist who wish to adopt the semantic approach to scientific knowledge. We need methods of generating or creating both scientific subjects and scientific predicates of scientific knowledge. We claim that inversion has a very crucial role to play in both kinds of constructions envisaged.
The distinctive nature of the questions to be addressed in the context of discovery have been partially indicated by Suppe, who clearly distinguished between two epistemologi- cally distinct stages in the process of theorization. One stage is the transition from phenomena to “hard” data about the physical system, and the second stage is the transition from the physical system to the postulates, etc., of the theory.
132 Chapter 5. Nature and Structure of Scientific Knowledge
The two sorts of moves are qualitatively different, the former being essentially
empirical or experimental—being in effect a “translation” from the phenomena to an idealized description of it in the vocabulary of the theory’s formalism, and
the latter being essentially mathematical or computational in nature.17
Suppe correctly identifies the first stage as one which involves counterfactuals, while the second stage involves mathematical computations (techniques). He further says, that the first stage is more complex than the second. Furthermore, it is a historical fact that the former is more difficult and more time consuming than the latter.
The problem of the construction of models will be taken up in the next chapter, where we will return to the issues identified here. The problem of constructing physical systems will be taken up in greater detail in the case studies in Part-III.
Before we get into the problem proper we shall critically discuss in detail the non-
statement view of Stegm¨uller on the structure of scientific theories.