| |
A philosophy of science that uses abstraction to identify the specific causal powers and liabilities of specific structures that are realized under specific conditions. This is a complicated sentence, but in order to understand its repeated emphasis on specificity — in sharp contrast to the philosophy of positivism — we need to begin by considering realism\'s view of causation.
Realism makes a fundamental distinction between:
{img src=show_image.php?name=2022.gif } the identification of causal mechanisms, which is typically the concern of so-called intensive research: the key question here is \'how does something happen?\'; and {img src=show_image.php?name=2022.gif } the identification of empirical regularities, which is typically the concern of so-called extensive research: the key question here is \'how widespread is something?\'Realism distinguishes between these two forms of research because \'what causes something to happen has nothing to do with the number of times it happens\' (Sayer, 1985a). Yet in the 1960s and 1970s much of mainstream human geography effectively erased this distinction. spatial science was preoccupied with the identification of empirical regularities, with a search for \'order\' and \'pattern\', and this tradition of extensive research was continued in the prediction of empirical regularities using space-time forecasting models. All of these projects were tacitly informed by the philosophy of positivism. But they faced a serious barrier, because the empirical regularities that they detected could only approach the status of the scientific laws envisaged by positivism — could only be regarded as laws of constant conjunction — provided both the mechanism capable of generating them and the conditions under which they were produced were held constant. These requirements can only be satisfied, however, if research is confined to a closed system. The problem, as Chouinard et al. (1984) explained, was this: all of the cases with which human geography and the other human and social sciences are centrally concerned are open systems.
On the basis of this equation between the identification of empirical regularities, the conduct of extensive research and the existence of closed systems, realism makes three critical interventions:
(a) The success of different sciences in identifying and predicting empirical regularities is determined in large measure by their ability to devise either empirically or (more usually) experimentally closed systems. In open systems it is not possible to guarantee the symmetry of explanation and prediction required by positivism. (Sayer, 1984a)
(b) The analysis of open systems requires the elaboration of a stratified and differentiated ontology, and it is here that realism makes its strongest and most strategic claim (Yeung, 1997). Both empiricism and positivism assume that the world is made up of events: these are the \'empirical particulars\' of science, and so their observation is accorded a special privilege as the leading edge of scientific discovery. In opposition to such \'atomism\', realism provides a multi-tiered ontology in which the world is made up not only of events but also of mechanisms and structures (see the figure). The connection between the last two is crucial. \'Structures\' are seen as sets of internal relations that have characteristic ways of acting: more formally, they possess \'causal powers and liabilities\', which are seen as constitutive because they possess these powers and liabilities by virtue of what they are. Precisely because these powers and liabilities enter fundamentally into the constitution of structures they are thus held to be necessary. Whether they are activated or not, however, is contingent: less technically, whether they are put into practice through the \'mechanisms\' of the structure (or not) depends on the circumstances. Seen like this, the task of a realist science is to tease out the causal chains that situate specific events within specific mechanisms and structures. The technical term for this procedure is the recovery of ontological depth: whereas positivism collapses the world into a single plane pockmarked by the space-time incidence of events, realism seeks to recover the connective tissue between the differentiated, foliated domains of events, mechanisms and structures.
{img src=show_image.php?name=bkhumgeofig64.gif }
realism and strategies for geographical inquiry (Gregory, 1985a, p.72)
(c) The identification and recovery of the various mechanisms and structures that make up the world is far from straightforward, because they are not immediately inscribed in the taken-for-granted categories which we draw upon in our everyday, \'common-sense\' discourse. Their disclosure thus requires a process of abstraction that is progressive, reflexive and essentially iterative (Yeung, 1997). In other words, realism relies on a research strategy in which theoretical categories inform and are in turn informed by empirical materials. Rather than privilege supposedly \'theory-free\' observations, realism places a premium on theoretical work: both on the critique of theoretical systems and on the co-determination of theoretical and empirical systems (Hesse, 1974; Gregory, 1978).
These three interventions can be used to underwrite both the natural sciences and the human and social sciences. Although physical geography did not engage with realism in any systematic fashion until the 1990s (see Richards, 1990a, 1990b), realist perspectives were opened up across the spectrum of the human and social sciences in the early 1980s: in history (e.g. McLennan 1981), in sociology (e.g. Keat and Urry, 1981), and in human geography (e.g. Williams, 1981; Sayer, 1982, 1985a; Chouinard et al., 1984). Within the human and social sciences the \'mechanisms\' that are the mainsprings of realism were usually identified with systems of social practices: indeed, Williams (1981) insisted that \'in a fundamental sense the concept of practice lies at the heart of the realist account\'. Many thinkers sought to establish filiations between realism and both historical materialism and structuration theory, and in doing so were led to a view of social practices as (i) being dependent upon knowledgeable and capable human subjects (although not reducible to them) and as (ii) having their effects determined in some substantial degree by contingent features of the settings in which they occur (Gregory, 1985). Double-edged claims of this sort sparked two major debates around realism in human geography.
(d) The appeal to \'knowledgeable and capable human subjects\' was intended to distance realism from essentialism (the belief that there is an essential reality lying \'behind\' the surface particulars of the world) and from structuralism (which displaces the human subject altogether). If this appeal is taken seriously, however, then it has to be allowed that the human and social sciences confront a world that is pre-interpreted, and that those interpretations are of basic importance to any explanatory account. As Keat and Urry (1981) recognized, this makes the abstractions of realism insufficient. For the fact is that most of us most of the time do not make sense of the world through the clean and clinical abstractions that lie at the centre of the realist account; our \'lay\' constructions must not be severed from the \'scientific\' explanations provided by realism. As Sayer (1985b) put it:
In real life we live in conjunctures whose boundaries are arbitrary; they haphazardly cut across structures and causal relations, and unless we devote considerable energy to their understanding, we only disentangle such conjunctures sufficiently for us to cope with everyday tasks. As theorists, however, we seek to understand the world by making rational abstractions which isolate unified objects, structures or groups, and we try to conduct concrete research by starting from such abstraction.Elsewhere Sayer (1984b) argued that the separation of these two facets, which he called the \'expressive\' and the \'objective\' respectively, had serious consequences for the social sciences and for social policy.
(e) The appeal to \'contingent features\' of the \'settings\' in which social practices occur — in other words, to their conditions of operation — was intended to prise realism away from any determinism. In contrast to positivism, realism regards scientific \'laws\' as statements of necessity not universality. Sayer illustrated the difference between the two by a simple thought experiment. Consider gunpowder: it has the (necessary) causal power to explode, yet it does not do so anywhere and everywhere. Whether it does so depends on (contingent) circumstances: it \'depends on it being in the right conditions — in the presence of a spark, etc. So although causal powers exist necessarily by virtue of the nature of the objects which possess them, it is contingent whether they are activated or exercised.\' By extension, too, their effects \'depend on the presence of certain contingently related conditions\'. What this means, Sayer concluded, is that in \'concrete research\' — in the examination of the exercise and effect of causal powers — space (or, more accurately, spatial configuration) \'makes a difference\':
In closed system … science the contingencies of spatial form are either rendered constant or a matter of indifference where they concern spatial relation between objects which do not causally interact. … In [open] systems we have a continually changing jumble of spatial relations, not all of them involving objects which are causally indifferent to one another. So even though concrete studies may not be interested in spatial form per se, it must be taken into account if the contingencies of the concrete and the differences they make to outcomes are to be understood. (Sayer 1984a; see also 1985b)This was an important conclusion for the human and social sciences as a whole, not just human geography, because it effectively undermined the possibility of any aspatial social science. But some critics regarded it as the weak version of a much stronger thesis. In their view spatial configurations are important not only for concrete research but also for \'abstract research\'. To use Sayer\'s own example: gunpowder is constituted as gunpowder, and hence possesses its specific causal powers, by virtue of the time-space relations which exist between its elements. This means that its constitution cannot be accounted for either by its time-space relations alone (the error of spatial science) or by its elements alone (the error of any compositional approach): they must be considered together. This turned out to be of decisive importance in providing an ontology for a socio-spatial science. As Urry (1985) put it, \'the social world should be seen as comprised of space-time entities having causal powers which may or may not be realised depending on the patterns of spatial-temporal interdependence [between them]\' (see also Bhaskar, 1986).
Realism was a powerful presence in human geography in the 1980s, but its star seemed to wane in the 1990s. In part, perhaps, this was a result of the connections made between realism and historical materialism and between realism and structuration theory. The retreat from (or advance beyond) these formulations seems to have gone hand in hand with a displacement of realism from the central position it had assumed within post-positivist human geography (cf. Harvey, 1987; Pratt, 1995). In part, perhaps, this was also the result of a profound uncertainty about how accounts conducted under the sign of realism were to be written. Here one needs to remember that \'realism\' refers not only to a twentieth-century philosophy but also to a mode of representation in the visual arts and literature which was particularly prominent in Europe and North America in the nineteenth century. This is not to say that realist philosophies require a realist aesthetics — they almost certainly do not — but simply to note that the attentiveness to theoretical work which realism succeeded in making so important for analysis during the 1980s was, in the next decade, extended to equally searching theoretical reflection on description (cf. Sayer, 1989). This \'crisis of representation\' called into question not only realist methodologies — the almost forensic precision of its \'rational abstractions\' was increasingly seen as problematic — but also, in the eyes of some critics, realist ontologies. Some forms of postmodernism made \'ontological depth\' yield to \'depthlessness\', for example, while the rise of post-structuralism induced considerable suspicion towards the sort of \'structures\' envisaged by realism and also fostered the development of a new \'analytics of the surface\' (see also non-representational theory). (DG)
References Bhaskar, R. 1975: A realist theory of science. Leeds: Leeds Books; reprinted 1978, Brighton: Harvester. Bhaksar, R. 1979: The possibility of naturalism: a philosophical critique of the contemporary human sciences. Brighton: Harvester. Bhaskar, R. 1986: Scientific realism and human emancipation. London: Verso. Chouinard, V., Fincher, R. and Webber, M. 1984: Empirical research in scientific human geography. Progress in Human Geography 8: 347-80. Gregory, D. 1978: Ideology, science and human geography. London: Hutchinson; New York: St. Martin\'s Press. Gregory, D. 1985: People, places and practices: the future of human geography. In R. King, ed., Geographical futures. Sheffield: Geographical Association. Harvey, D. 1987: Three myths in search of a reality in urban studies. Environment and Planning D: Society and Space 5: 367-76. Hesse, M. 1974: The structure of scientific inference. London: Macmillan; Berkeley, CA: University of California Press. Keat, R. and Urry, J. 1981: Social theory as science, 2nd edn. London: Routledge. McLennan, G. 1981: Marxism and the methodologies of history. London: Verso; New York: Schocken. Pratt, A. 1995: Putting critical realism to work: the practical implications for geographical research. Progress in Human Geography 19: 61-74. Richards, K.S. 1990a: \'Real\' geomorphology. Earth Surface Processes and Landforms 15: 195-7. Richards, K.S. 1990b: \'Real\' geomorphology revisited. Earth Surface Processes and Landforms 19: 277-81. Sayer, A. 1982: Explanation in economic geography: abstraction versus generalization. Progress in Human Geography 6: 66-88. Sayer, A. 1984a: Method in social science: a realist approach. London: Hutchinson (2nd edn. 1992, London: Routledge). Sayer, A. 1984b: Defining the urban. GeoJournal 9: 279-85. Sayer, A. 1985a: Realism in geography. In R.J. Johnston, ed., The future of geography. London: Methuen, 159-73. Sayer, A. 1985b: The difference that space makes. In D. Gregory and J. Urry, eds, Social relations and spatial structures. London: Macmillan, 49-65. Sayer, A. 1989: The \'new\' regional geography and problems of narrative. Environment and Planning D: Society and Space 7: 253-76. Urry, J. 1985: Social relations, space and time. In D. Gregory and J. Urry, eds, Social relations and spatial structures. London: Macmillan, 20-48. Williams, S. 1981: Realism, Marxism and human geography. Antipode 13 (2): 31-8. Yeung, H.W.-E. 1997: Critical realism and realist research in human geography: a method or a philosophy in search of a method? Progress in Human Geography 21: 51-74.
Suggested Reading Lawson, V. and Staeheli, L. 1990: Realism and the practice of geography. Professional Geographer 42: 13-19. Pratt (1995). Sayer (1984a) [1992]. Yeung (1997). |
|
