Truth as Warrant for Existence
The solution of our problem is implicit in the ‘bidirectionality’ that we have already stressed as included in the referential nature of truth, that is: if a proposition is true (for whatever reason it could be stated as true), its referents must exist (for the same reason) in the corresponding ontological region.
This, however, is just a hint, because for the moment we have conventionally restricted the notion of truth to single statements, and proposed a unique kind of criteria of truth and referentiality (i.e. the operations), that are appropriate only for immediate truth. Therefore, we should treat of the different methods for securing indirect truth to single propositions, that we could call synthetically inferential methods. They are widely examined in the standard literature of philosophy of science and need not be recalled here. Let us simply note that the majority of such methods is such that the conclusion of the inference is not endowed with absolute certainty. This, however, is only an epistemic (not an epistemological) condition that does not detract anything from the ontological commitment of the statements involved. This simply means that, if one is not certain about the truth of a statement, one is also not certain about the existence of the referents of this statement, but not that these referents do not exist because of this uncertainty. In addition, we know that such methods (especially those applied in the mature sciences) normally secure a significant degree of confidence in the truth of their conclusions, and this offers a (not absolute but very reasonable) warranty also for the existence of the corresponding referents.A further step consists in defining truth for a set of propositions and this can be done only if a certain structure is introduced into this set; the most spontaneous idea is to consider this structure as consisting in logical links for which elementary formal logic offers us a suitable display of patterns and rules with direct impact on the truth of the statements involved.
In particular, this is the conception underlying the sentential view of scientific theories that has dominated the analytical philosophy of science for most of the twentieth century and has promoted the nomo- logical-deductive model of scientific explanation. The typical schematization of this approach was that of considering a scientific theory as the logical conjunction of all its hypotheses (commonly presented as “laws”), and the testing of a theory as a logical deduction from these hypotheses (in conjunction with some empirically testable premises) of a singular empirically testable conclusion: if this conclusion were found to be true, the theory would be confirmed, if it were found to be false, the theory would be “falsified”. The trouble with this approach is that the confirmation is always not absolutely secured (for simple logical reasons), and that the falsification would concern the theory “as a whole” without indicating with certainty which one, among the various hypotheses, is false and makes false the logical conjunction making up the theory. These are the fundamental reasons that have shown the inadequacy of this approach to account for scientific change as it actually occurs and, especially after Kuhn's book on The Structure of Scientific Revolutions,[17] have started new trends in the philosophy of science.We are not interested in discussing these trends but we point out that they have in common the rejection of the statement view of theories and of the nomological-deductive model of scientific explanation. The first aspect almost automatically prevents the extension of the notion of truth to theories: if truth has been defined for single statements and then extended to logically structured sets of statements, but theories are not logically structured sets of statements, it follows that scientific theories are not the kind of entities to which the property of truth or falsity applies. What are then theories? The answers to this question are quite similar: theories are images, representations, models, global perspectives or Gelstalten which are put forward in order to understand and explain the data ascertained in a certain domain.
They perform this task by introducing theoretical entities that are the constituents of the representation or model but are not found in the data. It might be interesting to show how these characterizations tacitly rely upon that capability of intellectual intuition of which we have spoken above, but we want to consider other aspects of this issue.It is right to point out that we do not usually qualify as “true” an image, a representation, a model, but we would rather qualify them as “adequate”, “faithful”, “correct”, “appropriate”, “useful”, “reliable”. Nevertheless it is no less obvious that such models, or images, or representations are of no use from a cognitive and practical point of view unless they are expressed and made explicit by means of certain statements which may be far from expressing all the representational content of the intellectually conceived model, but have the great advantage of being communicable, true or false, and testable. For example, the map of a city is certainly far from being considered “true” in a strict sense, but it enables one to extract from it a certain number of true statements, like “the railways station is located at London Square”, “the distance between the Cathedral and your hotel is less than the distance from the City Park and the Modern Art Museum”, “you must walk to the South if you want to go from your hotel to China Town”, and so on. This simple example is sufficient to show that it is certainly wrong to maintain that a scientific theory is just a set of statements but it is equally right to maintain that it is also a set of suitably connected statements. This “suitably” has a complex meaning: from the one hand it indicates the logical correctness of the links, but on the other hand it alludes to the importance, to the “representational relevance” of the links.
Having (partially) redeemed the statement view of theories we are entitled to speak in an analogical sense also of the truth of a theory, considering the statements through which it is actually formulated.
By this we do not pretend, as van Fraassen would put it, that a scientific theory tells a “literally true story” about the world, but that it offers us a testable cognitive representation in which certain theoretical entities occur, that are elements of certain statements. We can maintain that the presence of these elements contributes to the understanding and explanation of our data not just because from the statements where they occur we can logically deduce the statements describing the data, but because, in addition, these theoretical elements appear to be causally related with the data in virtue of what is proposed in the theory, and this is a condition impossible to capture from a purely logical point of view. Therefore, the nomological-deductive model also retains a minimal value as a necessary condition for scientific explanation (in the sense that logical consistency and compatibility with the data is a necessary condition for fully accepting a theory) but they are not sufficient for deciding which part of the theory is responsible for a “falsifying” result, or whether this failure would simply require a “readjustment” rather than a total rejection of the theory. The “perspectivist” or “gestaltic” view of scientific objectivity that we have advocated offers a ground for debating these issues.[18] For the limited purpose of the present paper it is sufficient to stress that, by affirming a correctly understandable notion of truth for scientific theories, and recognizing that this truth entails the existence of the referents of all the concepts occurring in the true statements of the theory, we are entitled to say that also the theoretical entities so introduced exist with the same kind of existence of the other referents belonging to that ontological region.Let us note, in addition, that, usually, such theoretical entities are not practically accepted in mature sciences unless they are in a way “observed”. One must be aware, however, that the term “observation” is used here in a sense very different from the radically empiricist one that reduces it to the content of unaided sense perceptions.
Indeed, modern science relies upon instrumental observation, that avails itself of often very sophisticated instruments whose reliability and results (as we have already pointed out in Sect. 3) are granted by a robust application of scientific theories gradually accumulated up to the point of becoming reliable (within the scientific community) no less than sense perceptions in everyday life. Note that these instruments do not use theories, but the applications of theories, that is, the action of concretely existing things that exemplify, at different levels of complexity, the concepts of preexisting theories. For this reason technology is a very powerful warrant of scientific realism, both because it testifies of the capability to act of the referents of theoretical entities, and because it also enables scientists to “observe” even the unobservables.