The aim in this section is to consider what 'constructivism' means. It is instructive to begin this consideration with Leon J. Goldstein, who wrote prior to the narrativists and understood historiography as a specific form of constructivism:
The function of historical research is to constitute the historical past. As much as we want to say that a true account of some past event is true in virtue of the fact that it accords with what actually took place when the past was present, we have no way to make that belief operative in historical research. No examination of the actual character and procedures of historical study reveals a role for the real past to play, either in the formulation of historical hypotheses or in their confirmation. (1976, xix; my emphases)
First it seems that Goldstein makes a bold thesis of metaphysical constructivism ('constitution'), but then it becomes evident that Goldstein is actually, and somewhat confusingly, talking about an epis- temic problem. Indeed, later he refers to the role that historical assertions play in the 'verification' (1976, 3). His point is not to deny that there was a past, but that 'the past' as a metaphysical notion does not have a role in the historian's practice. Goldstein's position, and in his words 'the worst fears of the empiricist and the realist', is that 'we have no access to the historical past except through its constitution in historical research' (Goldstein 1976, xxi).
Goldstein's 'constitution of the past' ultimately boils down to the view that a historian has only inconclusive and incomplete evidence available out which to infer an account of history, or that the historian 'constructs a course of events which is supposed to make sense of what he has and knows' (1976, 15; original emphasis). This is achieved by 'an intellectual process of hypothetical reconstruction, or constitution, which in no way resembles perception' (Goldstein 1976, 15; similarly 58). This kind of constructivism appears to be a satisfactory description of how the historian arrives at his or her view. But it is ultimately unsatisfactory as a philosophical account because it does not address the question of whether the view so constructed could be true, in principle, of the 'metaphysical' past, assuming that we had no epistemic problems, as an omnipotent creature would not. Goldstein would probably accuse me of imposing a notion upon history which is brought 'from outside history, which history can never satisfy' (Goldstein 1976, 27). While I agree that the philosophy of historiography should be discipline-specific, I don't think that the application of philosophical notions in the context of historiography should be outlawed. It is true that one should not force historiography into any specific philosophical framework just because of the commonly used philosophical concepts and presuppositions (as in the case of relying on 'experience' as the foundation of knowledge). However, provided that this does not happen, any philosophical concept can be used to illuminate the discipline-specific questions and the nature of historiography. What Goldstein does not explicate is the fundamental problem with truth-claims in historiography: could the past in a metaphysical sense be structured like the output of the historian's inference, if we do not worry about the actual verification of the account?
Some have defined constructivism as a position according to which the objects of knowledge or facts are not independent of the knowing subjects and human activity (e.g. Kukla 2000, 19). This definition may initially look attractive with respect to some non-constructed object, such as a stone, for example, which appears independent of any knowing subject, in contrast to a painting, for instance. However, this definition is problematic even in the sciences as implied above. Some objects in science, such as certain rare elements and probably most synthesized objects in laboratories, simply are not independent of human activity. One might even say that their 'reality' requires construction. This might naturally be taken to mean that the application of constructivism is much wider than philosophers of science have assumed, which is arguably something that scholars in science studies would indeed claim. In the case of historiography, the situation is still different with regard to colligatory notions. Namely, the crux of the matter is not that their object of research is dependent on human intentional activity but that, strictly speaking, they do not have an individually specifiable research object (reference). That is the sense in which colligations are constructions.
The definition of non-constructed objects as being independent of the knowing subject above is problematic also for another reason. Even if an object of knowledge could be assumed to be independent of the knowing subject (e.g. gravity) this would not necessarily be decisive since it can hardly be maintained that knowledge of this object is independent of the knowing subject. It is my suggestion that constructivist talk applies best to the cognitive products of science and other fields. To say that a specific cognitive product, 'knowledge', is constructed is to say that it owes much to the subject-side and to claim that it is not constructed is to claim that it is in large part determined by the object-side. To put it in other words, the non-constructivist, and literally objectivist, position entails that knowledge merely reflects what there is. This kind of epistemological non-constructivism naturally entails that the object of knowledge exists independently of the knowing subject (although not necessarily independent of any knowing subject, since a previously ontologically constructed object can become an object of knowledge at a later stage). The constructivist position in turn says that the knowing subject determines knowledge, at least in part. We might then say that constructivism in historiography means that historiographical objects are dependent on the historian's activity - and thus on the subject- side. In other words, historiographical objects such as the 'Thaw' or Thompson's thesis concerning the birth of the English working class would not come into being without a historian's construction. It is necessary, again, to remark that constructivism in historiography does not need to include all kinds of historical objects. Finally, it remains to be said that constructed objects in this sense can be real if they are cognitively warranted. One should also remember that cognitive warrant implies adequate exemplification of data and minimal truth-functionality among the lower-level entities.