In his splendid Metaphysical Foundations of Modern Science, Burtt held that ‘[i]n the last analysis it is the ultimate picture which an age forms of the nature of its world that is its most fundamental possession. It is the final controlling factor in all thinking whatever’ (1932 : 3), an assumption which prompted him to search for the cosmology, or the world-view, underpinning modern science (that is, modern science up to 1905), and to contrast this modern vision to the medieval one. The resulting picture is most eloquent. The Middle Ages placed man at the centre of a finite universe and believed the whole world 'to be teleologically subordinate to him and his eternal destiny' (Burtt 1932 : 4), to such an extent that 'the categories in terms of which it was interpreted were not those of time, space, mass, energy, and the like ; but substance, essence, matter, form, quality, quantity…. Things that appeared different were different substances, such as ice, water, steam.' (p. 5) More specifically, their very understanding of space and time differed radically from ours, and reached back to Aristotle, for whom space was 'not something underlying all objects so far as they are extended, something occupied by them ; it [was] the boundary between any object and those which enclose it. The object itself was a qualitative substance rather than a geometrical thing.' (p. 84) As to time, Aristotle saw it as 'the continuous transformation of potentiality into actuality' (p. 85) ; the two categories of space and time, moreover, were completely lacking in importance.
From Kepler, but more especially from Galileo onward, this cosmology, or world-view, was completely transformed : man became an insignificant part of nature, separate from it and passively acted upon by the blind and mechanical laws of an infinite universe. First the scientists, then our whole scientific culture came 'to think about the universe in terms of atoms of matter in space and time instead of the scholastic categories' (Burtt 1932 : 17), and abandoned teleological explanations 'in favour of the notion that true explanations, of man and his mind as well as of other things, must be in terms of their simplest parts' (p. 16). The Aristotelian, qualitative notions of time and space, were replaced by quantitative ones :
- Physical space was assumed to be identical with the realm of geometry, and physical motion was acquiring the character of a pure mathematical concept. Hence in the metaphysics of Galileo space (or distance) and time become fundamental categories. The real world is the world of bodies in mathematically reducible motions, and this means that the real world is a world of bodies moving in space and time.
- (p. 83, italics original)
This story, retold countless times, has to my knowledge nowhere been followed up better than in Koyré's work (Koyré 1961, 1965, 1966a, 1966b, 1967). One of the greatest, if not the greatest, student of the scientific transformations from Galileo to Newton, Koyré did not hesitate to assert that this volte-face was above all a philosophical one (1961 : 252-69), by which I believe he meant something akin to Burtt's 'metaphysical foundations'. For Koyré, as for Burtt, these transformations amounted to no less than a revolution from the qualitative world of everyday experience and commonsense to the Archimedean world of reified geometry (Koyré 1961 : 261).
It is against this canvas that Koyré places Galileo's more theoretical achievements in dynamics, namely the fact that he intuited matter's inertial motion by imagining the motion of matter in a vacuum (a thought-experiment unthinkable in an Aristotelian Cosmos filled with matter) and derived the first law of dynamics. From a cosmological point of view, this notion of 'inertial motion' rested on no less than a complete inversion.
Indeed, Aristotle described a cosmos differentiated according to the intrinsic attributes of its elements (earth, water, air, fire and ether), one in which every element sought to be in its 'natural place', that is, in that location in space to which it belonged according to its natural attributes. Once in its natural place matter had no desire to move : rest, or absence of motion, was its natural state. Matter thus resisted movement, and objects moved only if a force overcame this resistance. With Galileo implicitly, and with Descartes explicitly, both motion and rest were conceived of as states of matter, states to which matter was indifferent ; in fact, even rest could be represented as a velocity (equal to zero), and Aristotle's question, 'What force can overcome matter's resistance to movement and create motion ?' was reformulated as 'What force can overcome matter's resistance to change its velocity and account for acceleration or deceleration ?'
When telling this saga Burtt and Koyré write of 'transformations', and Koyré more especially refers emphatically to the 'revolution' or the 'mutation' achieved by the New Science of the seventeenth century, thereby systematically opposing the 'evolutionists' in the interpretation of the history of science. Needless to say the camps are divided, and I quite openly side with the 'revolutionists' those who perceive discontinuities in the history of science.
I also wish to draw attention to the type of approach that Burtt and Koyré adopted. Burtt claims to be looking for the 'metaphysical foundations' of science, whereas Koyré speaks of its 'philosophical substructure'. As far as I am concerned such terms can lead to misapprehensions to readers unfamiliar with the history of science, and I wish to avoid them altogether. It is quite clear that both authors regarded their work as epistemological, and that their epistemological endeavours consisted in demonstrating that scientific revolutions do not result from a contest of theories : before new, revolutionary theories can be formulated, some deeper transformations have to take place. In some parts of their writings they describe these transformations as metaphysical or philosophical, but in the quotation opening this introduction Burtt writes of an underlying 'picture of the world' sometimes referred to as a cosmology, sometimes as a world-view. As to Koyré, he appears to be doing something very cognate. To avoid confusion let us therefore forget metaphysics and philosophy and suppose that, underneath any theory, there breathes a cosmology (or world-view).
Now, if the New Science of the seventeenth century accomplished a mutation of a cosmological nature, two conclusions necessarily follow : first, that theories nest in a cosmology from which they cannot be separated. Fully to understand and assess a theory from an epistemological point of view, one must start with a cosmological investigation. Second, if the Galilean science presupposed a cosmological revolution, that means that no science of dynamics (i.e., no set of coherent, quantifiable and falsifiable theories about movement) was possible within an Aristotelian world-view. In brief, some cosmologies preclude certain forms of theoretical discourses whereas others make them possible. Without a picture of the physical world enabling scientists to define space and time as we understand it, neither Galileo nor Newton could ever have formulated their celebrated theories, and a modern science of dynamics would have forever remained a purely chimerical project. Theories must be phrased in a language embedded in a cosmology and, from an epistemological point of view, they must first and above all be reinserted within this cosmology. On the two sides of a scientific revolution debates seem above all to be cosmological, not theoretical, although they are mostly phrased in theoretical terms.
Before spelling out my project, however, one more point must be stressed. Bunt and Koyré produced their masterly works without any explicit and coherent theory linking theories to cosmologies or metaphysical foundations. Similarly, I do not uphold any theory of the interaction between a given cosmos and language. I believe that theories and the concepts they are written in betray a vision of the world, and that some of those world-views thwart the emergence of a language designed for the rigorous description and analysis of some phenomena ; beyond that, I do not wish to get embroiled in debates on the causal connection between the two. Consequently, I posit axiomatically the existence of a cosmology behind any discourse aiming at explaining phenomena, and I further assume that this cosmos channels the manner in which the mind operates in drawing boundaries around the objects it observes.
It is within this particular tradition of the epistemology of science that I wish to elaborate this study of economics. Let us repeat our opening quotation by Burtt, with small addenda : 'in the last analysis it is the ultimate picture which an age forms of the nature of its [economic] world that is [economists] most fundamental possession. It is the final controlling factor in all [their] thinking whatever'. Like Burtt, I will try to unearth economics' 'metaphysical foundations', or to be more precise, to explore its cosmological underpinnings, and I will do so with a particular question in mind. In so far as I could ascertain from standard histories of economics, there are within economics three major research programmes, each one claiming to diverge completely from the others : Marxism, neoclassical economics and Keynesian economics. As to institutionalism, it is more of a residual category than a research programme with a hard core and auxiliary assumptions and, for many economists, it belongs more to sociology than pure economics. This is no value judgement, but simply a justification of my focus on constituted research programmes. I shall nevertheless reach out to institutionalism in the Conclusion.
This state of affairs cannot fail to raise some unsavoury questions. Are we to believe that all three research programmes are adequate tools to describe and analyse our economy, or that all are partially valid and complementing one another ? From the point of view of the history of science, either proposition is sorely problematic, Indeed, the history of science shows unequivocally that in most disciplines scientists originally held that different laws governed various subsets of the phenomena they studied, and that they explicitly held that a separate science must correspond to each subset. Until the Newtonian synthesis, for instance, students of nature thought that different laws governed the movement of bodies on earth and of those in space ; there were thus two sciences of dynamics. Similarly with chemistry ; until the synthesis of urea in 1828 chemists believed that different laws dictated the composition of inert and of organic matter, and that to those two types of matter corresponded two separate chemistries. And the same obtains with geology, not to mention the contemporary division of the world between quantum mechanics, the laws of which apply at the atomic level, and Einstein's theory of relativity, the laws of which rule the macroscopic universe. But contemporary physicists are hardly happy with this dichotomy and eagerly seek a unified theory.
In brief, if we have faith in the possibility of a scientific study of phenomena we should assume that the same research programme will ultimately account for all of them, instead of supposing that various research programmes are valid for different subsets (micro-economics would satisfactorily explain one type of economic phenomena, and macro-economics another ; or the economics of our own society would account for economic phenomena in a capitalist, monetary economy, but would be helpless when confronted with different economies, for example). If only at a linguistic level, one and the same set of concepts should suffice to describe and analyse all phenomena declared 'economic' in our own society.
If so, there is one alternative, and one only : either all contemporary economic research programmes fundamentally err, or one of the three should supplant the other two. This is the question I shall directly address, not as an economist but as an epistemologist. That is to say, I shall not broach economics directly through its theories but indirectly, through the cosmology within which these theories are embedded. As a result, my questions will differ markedly from those traditionally asked. Instead of emulating philosophers of science by asking, 'Are these theories scientific and in what way ?', I shall rather ponder, 'Do the cosmologies underlying economics' research programmes enable us to describe our economic world adequately ?' or, more precisely, 'Do they represent the economic cosmos in a way which makes it possible to develop the type of understanding they purport to achieve ?'
I therefore dissociate radically what I consider an epistemological enquiry from those disquisitions inspired by the philosophy of science. The latter are concerned with what constitutes a theory, and the relationship of theories to reality ; to that category of writings belong a host of books, reaching back to the nineteenth century (Blaug 1980 ; Boland 1982 ; Cairnes 1965 ; Coddington 1972 ; Friedman 1953 ; Green 1977 ; Hausman 1984 ; Hutchison 1965   ; Keynes, J. N. 1955  ; Lowe 1977 ; Machlup 1978 ; Robbins 1935 ; Robinson 1964 ; Rosenberg 1976 ; Samuelson 1948 ; Schoeffler 1955 ; Stewart 1979 ; von Mises 1978). Epistemological enquiries of the type attempted here, however, are extremely rare. I can only think of Ménard's masterly epistemological study of Cournot (1978). More indirectly, Fraser's Economic thought and Language (1937) is concerned with definitions and what lies behind them (in the same vein, I should also mention some of Machlup's articles, edited in Machlup 1963).
This divergence cannot be overemphasized. My project does not belong either to the philosophy or to the history of economics ; it aims at describing economic cosmologies and their relevance to the manner in which economics is practised now. This may appear questionable, because of the manner in which I have proceeded. Indeed, to unearth the cosmological assumptions of Newtonian science Burtt did not turn to the works of Lagrange and Laplace but to those of Galileo, Descartes, Newton. The reasons are simple. The works that initiate scientific revolutions make their cosmological premises quite explicit ; but soon, as Koyré observed, these premises are taken for granted, so much so that most practitioners lose awareness of the presuppositions of their own science and only the theories remain, often couched in the purest formalism. Therefore, even for economics as it is practised now, the cosmological clues are to be found in the initial statements.
For Marxian and Keynesian economics the choices of initial statements were self-evident, namely Marx's Capital and Keynes's General Theory. For neoclassical economics the decision was more difficult because of the apparent schisms separating the three dominant traditions, namely the Anglo-Saxon (following Jevons and Marshall), the Austrian (probably descended more from Wieser and von Mises than from Menger) and the one issued from Walras. I resolved this dilemma at the cost of an asymmetry in the presentation. Quite aware that contemporary developments from neoclassical economics derive more from the Austrian and especially the Walrasian traditions, but also quite alive to the fact that Marshall's Principles of Economics promised the richest cosmological harvest, I described the neoclassical cosmology mostly from secondary sources but referred to Marshall's Principles for specific examples while attempting, nonetheless, to connect Marshall's Principles to the other neoclassical traditions. I see no reason to be apologetic for this choice. Marshall seems to have reigned relatively unchallenged in the Anglo-Saxon world from the first edition of the Principles (1890) to that of Keynes's General Theory (1936) and even beyond (in 1952, Guillebaud declared that the Principles is still a standard textbook (Guillebaud 1952 : 186)). In the English-speaking world, its fame surpassed that of J.S. Mill's Principles of Political Economy and, according to Schumpeter, equalled that of Adam Smith's Wealth of Nations (Schumpeter 1954 : 830). Stigler deems it one of the two greatest books in the whole history of economic thought (Stigler 1962 : 223) and Deane declares that the 'neo-classical framework of analysis defined in Marshall's Principles of Economics established the main foundations of orthodox economic thought for roughly half a century after its publication' (1978 : 143). Shove puts it on a par with Smith's Wealth of Nations and Ricardo's Principles of Political Economy and Taxation (Shove 1942 : 147).
With Marx, and especially Keynes, however, I proceeded in the opposite fashion, building my argument directly from the Capital and the General Theory. Finally, I deliberately refrained from attempting to relate those treatises to the author's other publications or to other economic books and economic institutions of the time : hence the completely ahistorical nature of this work.
Needless to say, it was impossible to toy with cosmological issues without delving into the question of analogies ; in fact, as Ménard openly expressed and admirably demonstrated, analogies reveal better than anything else the underlying representation economists have of society what I have preferred to call their cosmology (Ménard 1978). Therefore, analogies lie at the very heart of this exploration and Mirowski's theses certainly could not be ignored (Mirowski 1991). Mirowski claims that neoclassical economics drew its main analogies from nineteenth-century proto-energetics. Unlike him, and despite the fact that I shall accept some of his theses although confining them to some later developments of neoclassical economics, I side with those who have espied economics' analogies in the science of Galileo and Newton in so far as the first generation of neoclassicals is concerned (Ménard 1978 ; Fisher 1986 ; Pribram 1986). I further recognize that various, and often conflicting, analogies happily cohabit within the same works and that these different analogies are often intertwined and as often misunderstood. I shall justify my stance as the analysis progresses and touch upon some of Mirowski's arguments when apposite. However, I believe Mirowski's interpretation of the history of science to be quite idiosyncratic, if not somewhat skewed, and I hold that it yields in places distorted views of the history of economics, especially when dealing with mercantilism and political economy. I address these more specific questions in Appendix 1, best consulted after having read Chapter 1.
Whether one sees the inspiration of economics in the science of Galileo and Newton, or in the field formalism of proto-energetics, however, both views beg the question, 'if neoclassical economics sought to emulate Galilean-Newtonian (or Maxwellian) science in the study of economic phenomena and has succeeded, shouldn't it then necessarily be the only valid research programme in economics, the others being but quackery, unless one assumes that Keynesian economics is to neoclassical economics what Einstein's physics is to Newton's ?' Some have dared draw such parallels, but most neoclassical and Keynesian economists shudder at the thought. The question, however, demands an answer. Through a careful scrutiny of neoclassical economics' cosmology I shall conclude that it never understood the basic premises of Galileo's and Newton's revolution and has remained a completely Aristotelian science ; the same applies to Marxism. Only Keynes's economics achieved a Galilean revolution, albeit one so tangled up in Aristotelian elements that it has never been so understood. Once properly appreciated, however, it can be shown to be the only research programme that can adequately describe our own economy, one that can easily be wedded to institutionalism while logically leading to the new probabilistic economics developed in the field of complexity theory, for instance. This is the book's central thesis.
Many of the ideas put forth in this essay are not new and lie, in one form or another, scattered through the literature. However, I hope in this volume to innovate, first in the manner of rearranging old ideas around a new question, and second, in spelling out here and there what has often remained elliptic.
 Dates in parentheses indicate the edition consulted, dates in square brackets indicate the original publication date. For ease of reference, both dates will be given at the first instance, thereafter only the date of the edition consulted will be given.