51. But what might these episodes be? And, in terms of our science fiction, how might our ancestors have come to recognize their existence? The answer to these questions is surprisingly straightforward, once the logical space of our discussion is enlarged to include a distinction, central to the philosophy of science, between the language of theory and the language of observation. Although this distinction is a familiar one, I shall take a few paragraphs to highlight those aspects of the distinction which are of greatest relevance to our problem.

    Informally, to construct a theory is, in its most developed or sophisticated form, to postulate a domain of entities which behave in certain ways set down by the fundamental principles of the theory, and to correlate -- perhaps, in a certain sense to identify -- complexes of these theoretical entities with certain non-theoretical objects or situations; that is to say, with objects or situations which are either matters of observable fact or, in principle at least, describable in observational terms. This "correlation" or "identification" of theoretical with observational states of affairs is a tentative one "until further notice," and amounts, so to speak, to erecting temporary bridges which permit the passage from sentences in observational discourse to sentences in the theory, and vice versa. Thus, for example, in the kinetic theory of gases, empirical statements of the form "Gas g at such and such a place and time has such and such a volume, pressure, and temperature" are correlated with theoretical statements specifying certain statistical measures of populations of molecules. These temporary bridges are so set up that inductively established laws pertaining to gases, formulated in the language of observable fact, are correlated with derived propositions or theorems in the language of the theory, and that no proposition in the theory is correlated with a falsified empirical generalization. Thus, a good theory (at least of the type we are considering) "explains" established empirical laws by deriving theoretical counterparts of these laws from a small set of postulates relating to unobserved entities.

    These remarks, of course, barely scratch the surface of the problem of the status of theories in scientific discourse. And no sooner have I made them, than I must hasten to qualify them -- almost beyond recognition. For while this by now classical account of the nature of theories (one of the earlier formulations of which is due to Norman Campbell,{18} and which is to be found more recently in the writings of Carnap,{19} Reichenbach,{20} Hempel,{21} and Braithwaite{22}) does throw light on the logical status of theories, it emphasizes certain features at the expense of others. By speaking of the construction of a theory as the elaboration of a postulate system which is tentatively correlated with observational discourse, it gives a highly artificial and unrealistic picture of what scientists have actually done in the process of constructing theories. I don't wish to deny that logically sophisticated scientists today might and perhaps, on occasion, do proceed in true logistical style. I do, however, wish to emphasize two points:

    (1) The first is that the fundamental assumptions of a theory are usually developed not by constructing uninterpreted calculi which might correlate in the desired manner with observational discourse, but rather by attempting to find a model, i.e. to describe a domain of familiar objects behaving in familiar ways such that we can see how the phenomena to be explained would arise if they consisted of this sort of thing. The essential thing about a model is that it is accompanied, so to speak, by a commentary which qualifies or limits -- but not precisely nor in all respects -- the analogy between the familiar objects and the entities which are being introduced by the theory. It is the descriptions of the fundamental ways in which the objects in the model domain, thus qualified, behave, which, transferred to the theoretical entities, correspond to the postulates of the logistical picture of theory construction.

    (2) But even more important for our purposes is the fact that the logistical picture of theory construction obscures the most important thing of all, namely that the process of devising "theoretical" explanations of observable phenomena did not spring full-blown from the head of modern science. In particular, it obscures the fact that not all common-sense inductive inferences are of the form

All observed A's have been B, therefore (probably) all A's are B.

or its statistical counterparts, and leads one mistakenly to suppose that so-called "hypothetic-deductive" explanation is limited to the sophisticated stages of science. The truth of the matter, as I shall shortly be illustrating, is that science is continuous with common sense, and the ways in which the scientist seeks to explain empirical phenomena are refinements of the ways in which plain men, however crudely and schematically, have attempted to understand their environment and their fellow men since the dawn of intelligence. It is this point which I wish to stress at the present time, for I am going to argue that the distinction between theoretical and observational discourse is involved in the logic of concepts pertaining to inner episodes. I say "involved in" for it would be paradoxical and, indeed, incorrect, to say that these concepts are theoretical concepts.

    52. Now I think it fair to say that some light has already been thrown on the expression "inner episodes"; for while it would indeed be a category mistake to suppose that the inflammability of a piece of wood is, so to speak, a hidden burning which becomes overt or manifest when the wood is placed on the fire, not all the unobservable episodes we suppose to go on in the world are the offspring of category mistakes. Clearly it is by no means an illegitimate use of "in" -- though it is a use which has its own logical grammar -- to say, for example, that "in" the air around us there are innumerable molecules which, in spite of the observable stodginess of the air, are participating in a veritable turmoil of episodes. Clearly, the sense in which these episodes are "in" the air is to be explicated in terms of the sense in which the air "is" a population of molecules, and this, in turn, in terms of the logic of the relation between theoretical and observational discourse.

    I shall have more to say on this topic in a moment. In the meantime, let us return to our mythical ancestors. It will not surprise my readers to learn that the second stage in the enrichment of their Rylean language is the addition of theoretical discourse. Thus we may suppose these language-using animals to elaborate, without methodological sophistication, crude, sketchy, and vague theories to explain why things which are similar in their observable properties differ in their causal properties, and things which are similar in their causal properties differ in their observable properties.