The philosophical literature on scientific explanation makes distinctions between (a) singular and theoretical explanation, and (b) complete and incomplete expla- nation. Singular explanations are explanations of singular facts or events (such as the fact that different ethnic groups are living in different parts of the city in Rotterdam).2 A singular explanation is expected to explain why a certain fact oc- curred in a certain way, to provide the causal history of it, or to inform us about the causal and structural relations that produced the explanandum phenomenon. If the phenomenon is fully accounted for, we have a complete singular explana- tion. A complete theoretical explanation, on the other hand, is supposed to provide all the causal and structural relationships that may explain all instances of the explanandum.
Most of the philosophical discussion is concerned with complete explana- tions:
If ‘scientific explanation’ does not mean ‘explanation actually offered in sci- ence’, the sense of the expression is far from obvious, and needs to be made clear. Many philosophers of explanation use it merely in the sense of ‘an ideally complete explanation’.
(Ruben 1990: 19)
Moreover, generally the focus is on complete singular explanations, rather than theoretical explanations. The rationale behind this is that what holds for complete singular explanations also applies to theoretical explanations (e.g. Ruben 1990: 19–23, Ylikoski 2001: 8), and that this abstract discussion is useful for it provides a reference point to evaluate actual explanations. The explanations actually of- fered by scientists differ more or less from these ideals; that is, they are usually incomplete. Obviously, the explanations we have encountered in this book are not complete explanations.
An important kind of incompleteness that Hempel discusses is partiality. An explanation is partial if it does not fully account for its explanandum:
Often, however, explanatory accounts exhibit a more serious kind of incom- pleteness.4 Here, the statements actually included in the explanans, even when supplemented by those which may reasonably be assumed to have been tacitly taken for granted in the given context, account for the specified explanandum only partially.
(Hempel 1965: 415)
For example, a complete singular explanation of the emergence of residential segregation in Rotterdam has to provide all the reasons why it emerged in Rotter- dam. It has to tell us about the arrival of foreign workers and people from former Dutch colonies; the initial housing decisions of the foreigners; Dutch housing policies; the employment areas for these people; their income levels compared to Dutch citizens; the immigration policies of the Dutch government; the racial preferences of the individuals; and all the events and facts that contributed to the emergence of residential segregation in Rotterdam. Obviously, such complete singular explanations are rare. Usually, one is content with providing the most important causal and structural factors. Hence, we usually face incomplete, or partial, singular explanations in real life. On the other hand, a complete theoretical explanation of the emergence of residential segregation should be able to account for all particular exemplifications of residential segregation in different times and places. Again, such complete theoretical explanations are rare, at least in social sciences. It is usually the case that general models account for some of the fac- tors that may explain the emergence of residential segregation. Actual theoretical explanations are usually partial.
We have seen that Menger and Schelling’s explanations are partial in the sense that they only take into account some factors that may explain the explanandum phenomenon and for this reason they cannot fully account for the explanandum phenomenon in Hempel’s sense. They ignore some of the factors known to be relevant for explaining the emergence of money and segregation (e.g. the fact that social and economic factors bring about segregation). Yet we should emphasise that partiality is not a special feature of explanations of unintended social phe- nomena; rather, most of the explanations offered by scientists are more or less partial:
many explanatory accounts offered in the literature of empirical science have the formal characteristics of partial explanations, and that as a consequence, they overstate the extent to which they explain a given phenomenon.
(Hempel 1965: 417)5
If we know that the explanans of a partial explanation is true, then having a partial explanation is better than having nothing – for we can develop the ex- planation by plugging in the other necessary elements. Yet some explanations may also fail to satisfy this truth condition, or the empirical adequacy condition. (Hempel and Oppenheim 1948: 248) The empirical adequacy condition states that ‘the sentences constituting the explanans must be true.’ This means that the statements of the antecedent conditions and the general regularities utilised by the explanation must be true. If this does not hold, then we have a potential explana- tion, instead of a true or correct explanation (Hempel 1965: 338). We have seen that this condition is not satisfied by the explanations we have examined in this book, because the descriptions of the antecedent conditions are rather conjectural or fictional. Particularly, Menger and Schelling’s explanations are potential (theo- retical) explanations in Hempel’s sense, for we cannot guarantee the truth of their premises.
We have noted that most of the philosophical literature on explanation is con- cerned with ideal explanations. In such a view, complete theoretical explanations are explanations of laws and generalisations with other laws and generalisations,6 and for this reason they are assumed to rest on an idea of well-established theory (e.g. Hempel 1965). A well-established theory may be defined as having all the conceptual (and methodological) resources for explaining the particular exem- plifications of the type of phenomenon it is concerned with. However, we have seen that Menger and Schelling’s explanations are not based on well-established theories, and they do not entirely rest on generally accepted generalisations and / or laws. Rather, these explanations are partial and based on partial models. Our examination of the contemporary literature on the emergence of a medium of exchange showed that partial models are important tools in the explanations of unintended consequences of human action, and that models may contribute in different ways to these explanations. The assumption of a well-established theory cannot help us understand how partial theoretical explanations work and how they could help us explain particular cases. A better understanding of the actual explanations of unintended social consequences rests on a better understanding of theoretical explanations and their relation to partial models.
Source: Aydinonat N. Emrah (2008), The Invisible Hand in Economics: How Economists Explain Unintended Social Consequences, Routledge; 1st edition.