In previous chapters, we have discussed two alternative (but complementary) in- terpretations of the invisible hand: the process interpretation and the end-state interpretation. Explanations that subscribe to the process interpretation of the in- visible hand may be considered as causal explanations for they focus on the mech- anisms that may bring about certain unintended social phenomena. The causal view of explanation7 suggests that an explanation has to inform us about the way in which entities in the real world are causally related to each other, or about the causes of the explanandum phenomenon (Salmon 1984, 1990).8 Yet a mere state- ment of the causes is not enough. A proper explanation has to inform us about the way in which causes are connected to their effects and to explicate the causal mechanisms that produce the phenomenon under study.9 Unfortunately, the causal view of explanation is mostly concerned with singular explanations; that is, with explanations of particular facts.10 For this reason, it does not help much to have a better understanding of the type of partial potential theoretical explanations we have encountered in this book. Given the requirements of the causal view, we may not argue that a proper explanation is provided unless a particular case is fully ex- plained and / or the existence of the proposed mechanisms is supported by ‘objec- tive evidence’ (Salmon 1998: 90). What we may take from the causal view is that the knowledge of (particular and / or general) causal mechanisms is important.
What about the end-state models? Since they do not seem to provide the causal mechanisms responsible for the origination of the end-state, they cannot be con- sidered as causal explanations – at least, not of the sort we have discussed above.11 But it may be that not all explanations are causal. Explaining some property of a matter A by referring to its structure can also be considered as an explanation. Since it is an open question whether structural explanations are causal or not, we need not constrain our view of explanation to the causal view or to a spe- cific type of causation. Rather, we may take it that a proper explanation should inform us about the structural and causal relationships that constitute or produce the phenomenon under consideration. If we accept this view we may consider the end-state models as informing us about the possible ways in which certain social phenomena may be constituted, thus providing certain bits of information that may help us in explaining particular cases of the social phenomenon under consideration (see Chapter 6). Although we will be talking mainly about the proc- ess interpretation, it seems wiser to have a view of explanation that is much more flexible than the causal view:
explanations work in virtue of something determining or being responsible for something. [. . .] we explain something by showing what makes it or what is responsible for it. The fault of causal theory of explanation was to overlook the fact that there are more ways of making something what it is or being responsible for it than by causing it.
(Ruben 1990: 231)
Mäki (1994: 159) proposes a flexible account of explanation that is consistent with this view, and which takes into account the fact that theories and explanations are, or may be, partial. He thinks that explanation can basically be characterised as re-description. According to this account, ‘explanation [. . .] involves rede- scription of explananda’ and re-description should be understood as re-description of what has been empirically described before (Mäki 1990a: 320). An important presumption of this view is the following:
it is primarily the task of scientific theory to do the explanatory work. Sin- gular phenomena cannot be explained by deriving their descriptions from empirical generalizations. [. . .] It is rather the case that theories account empirical facts directly. It is only by means of the conceptual resources of a theory – not being reducible to the observational language of empirical facts and generalizations – that empirical facts can be redescribed in a way which reveals what those facts are really are.
(Mäki 1990a: 321)
Here, Mäki is talking about singular explanations, and argues that by using the conceptual resources of a theory we may explain particular facts (i.e. provide singular explanations). Re-description of particular facts depends on theories. But this view also entails that theories and models represent parts of the real world. It is with the help of these representations that we are able to re-describe particular facts and provide singular explanations. Clearly, when we try to spell out the relation between theories, models that represent parts of the real world and re- descriptions (singular explanations) that rest on these representations, the mean- ing of ‘re-description’ gets blurred. Considering explanation as re-description is appealing, but what re-description really amounts to is not very clear. Consider Schelling’s chequerboard model, for example. Is the chequerboard model sup- posed to re-describe the way in which segregation emerges, or to represent the way in which it emerges and help us re-describe particular cases of segregation? But on what kind of prior empirical description of emergence of segregation does this re-description rest? A proper prior empirical description (whatever it really is) of emergence of segregation seems to be non-existent in Schelling’s case. Mäki may be right in that all explanations rest on the existing state of knowledge about the phenomenon under study, and this knowledge rests on empirical descriptions of that phenomenon. But characterising explanation as re-description does not tell much about the way in which the chequerboard model (or Menger’s model) is constructed and is supposed to explain. There seems to be a more complex relationship between theories, models and explanation – at least for the cases we have examined in this book.
Moreover, the term ‘re-description’ wrongly suggests that models represent particular phenomena. For example, concerning the chequerboard model, Sugden challenges the re-description view (particularly Mäki’s account of isolation and representation) by arguing that the chequerboard model does not represent any particular city:
it does not seem right to say that the chequerboard model isolates some aspects of real cities by sealing of various other factors which operate in reality: just what do we have to seal off to make a real city – say Norwich – become like a checkerboard? Notice that, in order to arrive at the checkerboard plan, it is not enough just to suppose that all locations are identical with one another (that is, to use a ‘generic’ concept of location): we need to use a particular form of generic location. So, I suggest, it is more natural to say that the checkerboard plan is something that Schelling has constructed for himself. If we think that Schelling’s results are sufficiently robust to changes in the checkerboard assumption, that assumption may be justified even if it is not an isolation.
(Sugden 2000: 22)
Sugden is right in that the chequerboard model is constructed and is not a model of a particular city. Yet this does not contradict with the statement that models represent and are isolated from other complexities of the real world. It is probably the terminology of the explanation-as-re-description view that wrongly suggests that models represent particular phenomena. As will be seen in the rest of this chapter, the relation between models, theories and explanations is complex. For this reason, we need a better picture of how models and theories represent and help us to explain.
To prevent other possible misunderstandings caused by the notion of ‘re-description’, and to mediate the complexity of the act of representation and explaining, we may argue that explanation entails discovering ways to conceptualise the phenomenon at stake:
Whether a fact as normally understood explains or is explained depends at least in part on the way in which the properties involved are conceptualised: relative to the conceptualisation of a property in one way, the fact may be explanatory, relative to a different conceptualisation of the same property, the fact is not explanatory.
(Ruben 1990: 177)
Explaining a complex phenomenon entails discovering how to conceptualise it, that is, to discover the way in which the world works, or may be working in producing a certain phenomenon. Given the importance of partial models in explanations of unintended social phenomena, and the conjectural character of the reasoning involved, we may further argue that novel partial potential theoreti- cal explanations (e.g. such as Schelling’s) involve ‘creative conceptualisations’ that are partly discovered through the construction of models and conveyed by way of presenting them. It is on this complex task of representation that explana- tion rests. The next section examines the way in which models represent.
Source: Aydinonat N. Emrah (2008), The Invisible Hand in Economics: How Economists Explain Unintended Social Consequences, Routledge; 1st edition.