PLEASE SCROLL DOWN FOR ARTICLE

This article was downloaded by: [York University Libraries]On: 22 March 2010Access details: Access Details: [subscription number 791654236]Publisher RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

International Studies in the Philosophy of SciencePublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t713427740

Gaston Bachelard's philosophy of scienceGary Cutting a

a Professor of Philosophy, University of Notre Dame, Indiana, USA

To cite this Article Cutting, Gary(1987) 'Gaston Bachelard's philosophy of science', International Studies in the Philosophyof Science, 2: 1, 55 — 71To link to this Article: DOI: 10.1080/02698598708573302URL: http://dx.doi.org/10.1080/02698598708573302

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial orsystematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply ordistribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae and drug dosesshould be independently verified with primary sources. The publisher shall not be liable for any loss,actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directlyor indirectly in connection with or arising out of the use of this material.

Gaston Bachelard'sphilosophy of science

Gary Cutting

The work of Gaston Bachelard on the philosophy of science (developed ina series of books published from 1927 to 1953) is well-known and extremelyinfluential in France. It is essential background for an understanding of thework of philosophers and historians such as Georges Canguilhem, LouisAlthusser, and Michel Foucault. Moreover, Bachelard's philosophy ofscience deals in an extremely interesting way with topics - particularlyconceptual change and scientific realism - that later became central for theanalytic philosophy of science dominant in English-speaking countries.Unfortunately, this important work is very little known outside of France.1

In this paper, I present the main lines of Bachelard's philosophy of science,emphasizing points of contact with recent Anglo-American work, andmake some brief evaluative comments.

Reason and Science

As Foucault once pointed out (1985, 5-6), Bachelard is part of aspecifically French tradition (beginning with Comte) that approaches theKantian project of a critique of reason through the history of science. (Bycontrast, Foucault notes, in Germany the approach, from Marx throughthe Frankfurt School, has been through historical reflection on social andpolitical structures.) According to Bachelard, reason is best known byreflection on science, and science is best known by reflection on its history.The first thesis derives from his conviction that the structures of reason areapparent not in abstract principles but in the concrete employments ofreason. Norms of rationality are constituted in the very process of applyingour thoughts to particular problems, and science has been the primarylocus of success in such applications. The proof of the second thesis - thatscience is best known through its history - lies in the repeated refutation ofa priori philosophical ideals of rationality by historical scientific develop-ments. Descartes, for example, held that rational science must be

International Studies in the Philosophy of Science Volume 2 Number 1 Autumn 1987© R.K.P. 1987 0269-8595/87/0201-0001 $2

55

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gary Gutting

grounded in clear and distinct intuitions of the essential properties ofmatter. This view is refuted by the fact that matter, as described bytwentieth-century physics and chemistry, is simply not available to ourintellectual intuition. We know it only through the indirections ofhypothetico-deductive inference from data that are themselves mediatedby complex instruments (1934, 138ff and Chapter 6). Similarly, Kant'sformulation of a transcendental, a priori analytic of principles thatregulate all employments of reason collapsed with the triumph of theories(relatjvity and quantum mechanics) based on the denial of such Kantianprinciples as the permanence of substance, which require a continuity ofenergy inconsistent with quantization (1940, Chapter 3). What initiallyseem to be a priori constraints on thought as such turn out to be contingentconditions derived from philosophers' inability to think beyond theframework of present science.

There are, then, no viable accounts of rationality except those derivedfrom the historical developments of scientific reason. To understandreason, philosophy must 'go to the school of science'. Here, as elsewhere(e.g., the development of metaphysical theories), the achievements ofscience are the dynamics behind all philosophical understanding. 'Sciencein effect creates philosophy' (1934, 3).

The rationality that philosophy tries to discover in the history of scienceis no more fixed and monolithic than that history itself. As we shall seeshortly, Bachelard finds sharp breaks in the history of science andcorresponding changes in the conception of reason. Moreover, Bachelardreminds us that there is, strictly speaking, no such thing as the history ofscience, only various histories of different regions of scientific work.Correspondingly, philosophy cannot hope to uncover a single, unifiedconception of rationality from its reflection on the history of science; it willfind only various 'regions of rationality,' ('les regions rationelles') (1949,Chapter 7). Bachelard, for example, analyses (in Le rationalisme applique)the rationalities implicit in nineteenth-century theories of electricity and ofmechanics. He agrees that the history of science tends to the integration ofdiverse regions of rationality but sees no place for a 'science in general' towhich would correspond a 'general rationality.' Bachelard was particularlyconcerned with a new rationality that he saw in the achievements oftwentieth-century physical science, especially relativity theory and thephysics and chemistry of quanta.

Bachelard's Model of Scientific Change

Because of his demand that the philosopher of science work from thehistorical development of the sciences, the center of Bachelard's philosophyof science is his model of scientific change. This model, which also provides

56

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gaston Bachelard's philosophy of science

his account of the nature of scientific progress, is built around four keyepistemological categories: epistemological breaks, epistemologicalobstacles, epistemological profiles, and epistemological acts.

Bachelard employs the concept of an epistemological break in twocontexts. First, he uses it to characterize the way in which scientificknowledge splits off from and even contradicts common-sense experiencesand beliefs. This sense of 'break' is fundamental for Bachelard, since itconstitutes science as a distinctive cognitive realm: '. . . scientific progressalways reveals a break [rupture], constant breaks, between ordinary[commune] knowledge and scientific knowledge' (1953, 207). Bachelardillustrates this claim with several examples that we can use to elucidate thekey features of epistemological breaks. He finds one simple example in achemistry text's comment that glass is very similar to wurtzite (zinc sulfite).The comparison is one that would never occur to common sense, since it isnot based on any overt resemblance of the two substances but on the factthat they have analogous crystalline structures. Thus, science breaks withordinary experience by placing the objects of experience under newcategories that reveal properties and relations not available to ordinarysense perception.

But we should not think of scientific breaks as merely a matter ofdiscovering new aspects of ordinary objects, of taking up where everydayexperience leaves off, as a telescope reveals stars not visible to the nakedeye. New, scientific concepts are required to give an adequate account ofeven familiar facts. This is very nicely illustrated by the case of Lamarck'sfutile efforts to use his exceptional observational abilities to develop anaccount of combustion in opposition to Lavoisier's. His approach was tonote carefully the sequence of color changes a piece of white paperundergoes when burned. On the basis of such observations, Lamarckinterpreted combustion as a process whereby the 'violence' of the fire'unmasks' the fundamental, underlying color of the paper (black)by stripping away successive chromatic layers. Bachelard argues thatLamarck's idea here is not merely wrong in the ordinary way of anincorrect scientific hypothesis. Rather, it is essentially anachronisticbecause it is based on immediate phenomenal experiences that Lavoisierhad already shown to be inadequate for the task of understandingcombustion. 'The time for direct, natural observation in the realm ofchemistry had passed' (1953, 219).

A final example shows how science may break with common-senseeven when employing models based on its language and concepts. This isthe case of Bohr's 'water-drop' model of the atomic nucleus. Via thismodel, Bohr pictured the protons and neutrons of the nucleus as forming adrop of water, the 'temperature' (internal energy) of which increased whena neutron was added and which partially 'evaporated' when a particle wasemitted from the nucleus. This model was an excellent aid to understanding

57

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gary Gutting

the process of fission, but its use of ordinary concepts must not mislead us.As Bachelard puts it, such words as 'water-drop', 'temperature', and'evaporation' occur only in quotation marks. In fact, the words are tacitlyredefined so that they express concepts that 'are totally different . . . fromthe concepts of common knowledge' (1953,216). (Imagine, Bachelard says,the stupidity of asking a physicist to make a thermometer to measure the'temperature' of the nucleus.)

This last example also illustrates Bachelard's second sort of epistemo-logical breaks: those that occur between two scientific conceptualizations.If nuclear 'temperature' is a very different concept from ordinaryphenomenal temperature, it is likewise very different from the classicalconception of temperature as the mean kinetic energy of a collection ofmolecules. This illustrates how science develops not only by breaks withordinary experience but also by breaks with previous scientific theories.For Bachelard, the most striking and important such breaks came withrelativity and quantum theory, which he saw as initiating a 'new scientificspirit.' This 'new spirit' involved not only radically new conceptions ofnature but also new conceptions of scientific method (e.g., new criteria ofexplanatory adequacy). Bachelard's detailed treatments of this topic (in,e.g., La valeur inductive de la relativite and in Le nouvel esprit scientifique)preceded by two or three decades similar discussions by Anglo-Americanhistorians and philosophers of science such as Kuhn and Feyerabend.

The language of epistemological 'breaks' suggests that there issomething to be broken, a barrier that must be shattered. Bachelardfollows out this suggestion with his notion of an epistemological obstacle.An epistemological obstacle is any concept or method that prevents anepistemological break. Obstacles are residues from previous ways ofthinking that, whatever value they may have had in the past, have begun toblock the path of inquiry. Common-sense is, of course, a major source ofepistemological obstacles. Thus, the animism of primitive common-sense,which inclined people to explain the world on analogy with vital processes(sex, digestion, etc.) was an obstacle to the development of a mechanisticphysics. Likewise, the still strong common-sense idea that phenomenamust be the attributes of an underlying substance blocked the rejection ofthe ether as the locus of electro-magnetic waves. More generally,Bachelard regards the common-sense mind's reliance on images as abreeding ground for epistemological obstacles. Images may have heuristicuse in science, but they have no explanatory force; and, if they do their jobproperly, they are eventually eliminated from scientific thought. Thus, ofBohr's planetary model of the atom, Bachelard says: 'The diagram of theatom provided by Bohr . . . has . . . acted as a good image: there isnothing left of it' (1940,119). But epistemological obstacles may also arisefrom successful scientific work that has outlived its value. The most strikingcases occur when the concepts and principles of an established theory lead

58

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gaston Bachelard's philosophy of science

us to regard new proposals as obviously absurd; e.g., the counter-intuitivefeel of quantum mechanics' rejection of classical determinism. Butpreviously successful scientific methods can also become epistemologicalobstacles. For example, the emphasis on direct observation that led in theseventeeth century to major breaks with Aristotelian science became anobstacle to eighteenth-century developments of atomic theories. Finally,traditional philosophy, with its tendency to canonize as necessary truthsthe contingent features of one historical period of thought, is anothermajor source of epistemological obstacles.

The views and attitudes that constitute epistemological obstacles areoften not explicitly formulated by those they constrain but rather operateat the level of implicit assumptions or cognitive and perceptual habits.Consequently, Bachelard proposed to develop a set of techniques designedto bring them to our full reflective awareness. He spoke of thesetechniques as effecting a 'psychoanalysis' of reason. Bachelard's use of thisterm signals his aim of unearthing unconscious or semi-conscious structuresof thought, but it does not express a commitment to the details of Freudiantheory.

Closely related to the concepts of epistemological obstacles andpsychoanalysis of reason is the idea of an epistemological profile. This is ananalysis of a given individual's understanding of a scientific concept, ananalysis that reveals the degree to which the understanding involveselements from various stages in the concept's historical development.These stages correspond primarily to various philosophical embodimentsof past scientific ideas (as well as to various common-sense notions). Thus,Bachelard (1940, 36-8) provides the profile of his own concept of mass,which he finds to be dominated by the classical rationalist conception (massas the primitive term of eighteenth-century mechanics). It also has a verystrong empiricist or positivist component (mass defined operationally aswhat is measured on scales). There are smaller, but still significant,contributions corresponding to the child's naive understanding of mass as adesirable quantity of matter and of the highly abstract and rationalizedconceptions of relativity theory and quantum mechanics. A parallel profileof Bachelard's understanding of energy (1940, 38-43) shows a similardomination by the rationalist conception of the eighteenth century but lessinfluence of the empiricist conception and more of the child's naive notion.

Obviously, an epistemological profile provides a record of theepistemological obstacles hindering the scientific thought of a givenindividual. But Bachelard does not make the scientistic move of rejectingall positive significance for those elements of the profile that are not fullyadequate to the achievements of contemporary science. Rather, particularlyin his later work, he comes to see the full range of the elements ofepistemological profiles - including the philosophical and common-sensecomponents - as valid on various levels of human life and experience.

59

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gary Gutting

Scientific inadequacy does not entail complete invalidity. Instead,Bachelard concludes that the complexity of an epistemological profileshows that 'a single philosophy cannot explain everything' and that 'it isnecessary to group all the philosophies to obtain the complete notionalspectrum of a particular piece of knowledge' (1940, 42). Bachelard's fullappreciation of the nonscientific ('poetic') dimension of human experienceis developed in a series of books beginning with The Psychoanalysis of Fire.On one level, this book simply provides a detailed analysis of theepistemological obstacles connected with our images of fire. (One exampleis the 'violence of fire' that we noted in Lamarck's thinking aboutcombustion.) But Bachelard's reflections lead him to an appreciation ofpoetic images and experience for their own sakes. He continues to resist'the ontological temptation of beauty' (Smith, 1982, 77): that is, he issteadfast in rejecting any view ('naive realism') that would make thecontents of ordinary, subjective experience as or more real than scientificobjects. But he sees the subjective realm that feeds the poetic spirit asintrinsically valuable and a necessary complement to scientific knowledge.

The concept of an epistemological act counterbalances that of anepistemological obstacle. Whereas epistemological obstacles impede scien-tific progress through the inertia of old ideas, 'the notion of epistemologicalacts corresponds to the leaps [saccades] of scientific genius that introduceunexpected impulses into the course of scientific development' (1951, 25).An epistemological act is not, however, just a change; it has a positivevalue that represents an improvement in our scientific accounts. There are,accordingly, different values that must be accorded to different episodes inthe history of science. Consequently, Bachelard holds that writing historyof science is different from writing political or social history. In the lattercase, 'the ideal is, rightly, an objective narration of the facts. This idealrequires that the historian not judge; and, if the historian imparts the valuesof his own time in order to assess the values of a past time, then we areright to accuse him of accepting "the myth of progress" ' (1951, 24). But inthe case of the history of the natural sciences, progress is no myth. Presentscience represents an unquestionable advance over its past, and it isentirely appropriate for the historian of science to use the standards andvalues of the present to judge the past. Application of these standardsresults in a sharp division of the scientific past into Thistoire perimee' (thehistory of 'outdated' science) and Thistoire sanctioned' (the history ofscience judged valid by current standards). More broadly, following F. K.Richtmyer, Bachelard distinguishes between the story of science - anaccount of past scientific achievements that have contributed to ourpresent body of knowledge - and mere history of science, which includesefforts that have no positive place in the genealogy of current science(1951, 27). Bachelard also speaks of an account of science that 'starts fromthe certainties of the present and discovers in the past progressive

60

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gaston Bachelard's philosophy of science

formations of the truth' as 'recurrent history' {Vhistoire recurrente).Georges Canguilhem points out (1970, 197-8; 1978, 21ff) that this

Bachelardian writing of the history of the past on the basis of the present isnot equivalent to the now generally disdained 'Whiggish' approach to thehistory of science. For one thing, Bachelardian history does not try tounderstand past science in terms of present concepts. It realizes the need toexplicate the past in its own terms. For another, there is no assumption ofthe immutable adequacy of present science. Precisely because they arescientific, the present achievements by which we evaluate the past maythemselves be surpassed or corrected by future scientific development.Our evaluation of the past in terms of the present is, in Canguilhem'swords, not the application of 'a universal touchstone' but 'a selectiveprojection of light on the past' (1977, 22).

But, for Bachelard, even though all scientific results are open torevision and some can be definitively rejected, others must be accepted aspermanently valid achievements. Thus, he says that phlogiston theory is'outdated [perimee] because it rests on a fundamental error.' Historianswho deal with it are working 'in the paleontology of a vanished scientificspirit' (1951, 25). By contrast, Black's work on caloric, even though mostof it has long been jettisoned, did yield the permanent achievement of theconcept of specific heat. 'The notion of specific heat — we can assert withequanimity - is a notion that is forever a scientific notion . . . One maysmile at the dogmatism of a rationalist philosopher who writes 'forever'regarding a scholastic truth. But there are concepts so indispensable in ascientific culture that we cannot conceive being led to abandon them' (1951,26).

How is this idea of unalterable progress consistent with Bachelard'sinsistence that all scientific results are open to revision? How can anachievement be 'permanent' and at the same time open to correction in thewake of an epistemological break? Bachelard's response is that anepistemological break is not merely the rejection of past science but also apreservation, via reformulation, of old ideas in a new and broader contextof thought. Specifically, past results are replaced by generalizations thatreject them as unconditionally correct but preserve them as correct undercertain restricted conditions. Bachelard finds a model here in thedevelopment of nonEuclidean geometry. This development refutes theclaim that the Euclidean postulates express the sole truth about geometrybut at the same time presents these postulates as defining one exemplifica-tion of a more general class of geometries (i.e., Euclidean geometry is theparticular geometry possessed by a space of zero curvature). In the sameway, 'Newton's astronomy can . . . be seen to be a special case ofEinstein's "pan-astronomy" ' (1934, 45). This is so not merely because, toa certain approximation, Newtonian calculations yield the same numbersas Einsteinian calculations but because key Newtonian concepts such as

61

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gary Gutting

mass and velocity can be shown to be special simple cases of thecorresponding Einsteinian concepts. Bachelard characterizes this processof replacement by generalization as 'dialectical,' not in the Hegelian senseof a synthesis of opposites but in the sense of a process of conceptualexpansion whereby what previously appeared to be contraries (e.g.,Euclidean and Lobachevskian geometries) are seen as complementarypossibilities. Earlier concepts are not mysteriously 'sublated' into a higherunity but are rectified (corrected) on the basis of superior successorconcepts that allow us to explain precisely the extent to which they areapplicable.

This account of scientific change allows Bachelard to reject thecontinuity of science and still accept its progress. Science develops by aseries of epistemological breaks that make it impossible to regard itshistory as a linear accumulation of truths within a single conceptualframework. The conceptual framework of science at one stage will berejected as erroneous at later stages. Nonetheless, some of its results maybe permanent scientific achievements in the sense that they will bepreserved as special cases within all subsequent scientific frameworks.Each successive framework will represent progress over its predecessors inthe sense that it has a more general perspective from which the range ofvalidity of previous perspectives can be assessed.

Bachelard's picture of science as discontinuous but progressive involvesan essential role for two complementary factors that are usually excludedfrom the history of science: errors and norms. The former are generallythought of as unfortunate, contingent deviations from the path of science,due to scientifically extraneous factors such as prejudice, inattention, andignorance. Strictly speaking, the scientist as such (like the mathematicianof Republic I) does not make mistakes. The history of science is the historyof scientific truth, in which errors have no essential role. For Bachelard,however, even a permanent scientific achievement such as Newton'smechanics may be judged an error (i.e., in need of correction) from themore general viewpoint of a later theory. Such errors are essential stages inthe development of science and are, as we have seen, preserved in rectifiedform by subsequent theories. Similarly, scientific norms are generallyregarded as outside the historical process of scientific development, sincethey are thought of as atemporal, universally valid principles of reason.Bachelard however sees norms as formed in the very process of thehistorical development of science. Norms are products of science's rationalactivity and can be superseded at later times by other norms that provethemselves more adequate. In this way, Bachelard's model of scientificchange fulfills his program of treating reason as a genuinely historicalphenomenon.

Bachelard's approach to scientific change promises solutions to some ofthe fundamental problems of post-Kuhnian philosophy of science. For

62

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gaston Bachelard's philosophy of science

example, by suggesting that we can maintain the progressiveness of sciencewhile denying the continuity of its development, he offers a way ofreconciling historical evidence of radical shifts in paradigm with therationality of science. Similarly, his treatment of norms suggests that thisrationality can itself be regarded as fundamentally historical. However,Bachelard's account of these matters needs further development, particu-larly with regard to the 'dialectical' process whereby past scientificconcepts emerge as special cases of present scientific concepts. In preciselywhat sense, for example, is Newtonian mass a 'special case' of relativisticmass? More generally, what are the general conditions under which oneconcept may be said to be a special case of another? A plausible account ofthese matters requires a more careful analysis of the nature of concepts andof the meanings they express than Bachelard offers.

The epistemological and metaphysical ramifications of Bachelard'smodel of scientific change

Bachelard's philosophy of scientific change involves, via its presuppositionsand consequences, more general epistemological and metaphysical I posi-tions. These positions derive from his reflections on the methodologicaland ontological results of actual scientific practice. Bachelard dismisses asgroundless speculation any purely philosophical theories about how weknow and what there is. The nature of knowledge and of reality can beknown only by reflection on the successful applications of reason to theunderstanding of our world. (This view of epistemology and metaphysicsis, of course, just a generalization of the considerations noted above thatled Bachelard to insist on approaching the philosophy of science throughthe history of science.) It is important to realize that Bachelard'ssubordination of philosophy to science is not an instance of positivisticscientism. He does not think - as does, e.g., Quine with his notion ofnaturalized epistemology - that philosophy itself is part of science.Philosophy for Bachelard is a reflection on sciences but its methods andresults do not share the empirical character of scientific disciplines.

What, in any case, does Bachelard's reflection on science yield as to thegeneral nature of knowledge? Where does he stand as an epistemologist?Since he explicitly proclaims his epistemology to be nonCartesian, we canprofitably proceed by contrasting his position with that of Descartes.

Bachelard accepts Descartes' view that knowledge arises from acritically methodical questioning of accepted beliefs, but he rejectsDescartes' claim that to attain knowledge this questioning must terminatein indubitable intuitive certainties (cf. Tiles, 1984, 28-33). Cartesianfoundationalism is entirely implausible relative to Bachelard's view ofscience's history as a series of epistemological breaks. What reason do we

63

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gary Gutting

have to think that the basis of our science will prove any firmer than didthat of our forebearers? But, beyond this, Bachelard argues that there areno grounds for according Cartesian intellectual intuitions (or any other sortof claim to direct intuitive knowledge) the privileged certainty thatDescartes does. His critique of intuition centers on the two poles of anintuitive experience: its subject and its object.

With respect to the latter, the object of a foundational intuition must beanalyzable into a set of simple elements, each known fully and unambigu-ously. To the extent that the objects of our experience contain hiddencomplexities, our judgments about them are subject to correction in thelight of more penetrating analyses. This is why Descartes, in particular,required that clear and distinct perceptions effect a reduction of theirobjects to 'simple natures'. (Similarly, foundationalists in the modernempiricist tradition take unanalyzable sense data as the ultimate objects ofexperience.) Bachelard's criticism of this aspect of foundationalism is thateven the most apparently simple objects of our intuition have later provedto have complex hidden structures. Their apparent simplicity was reallyonly an ultimately misleading simplification. Specifically, Bachelard seesmajor breaks in the history of science as due to the discovery of hiddencomplexities in objects that had been regarded as simple intuitive givens.One famous example is Einstein's analysis of time and simultaneity.Another less known but very striking example is the twentieth-centurystudy of hydrogen through the analysis of its atomic spectrum. (Cf. 1934,148ff.) Here physicists did at first proceed in accord with the Cartesianideal. They began with the simple case of the hydrogen atom (a singleproton orbited by a single electron) and tried to show how the empiricalformula that described its spectrum (the Balmer formula) could begeneralized to arrive at formulae describing the spectra of more complexatoms. In this way, it seemed that knowledge of complex cases could bedeveloped from knowledge of the simple case. But in fact the developmentwas in exactly the opposite direction. 'In order to give a detailed account ofthe spectroscopic data, the more complicated spectrum (here that of thealkaline metals) had to be treated first' (1934, 154-5). For example, thefine structure of spectral lines (i.e., the doubling noted in what at first seemto be single lines) was found first in the spectra of more complex atoms;and, only because it was noted there, did scientists look for it in thespectrum of hydrogen. Similarly, the role of the angular momentum of thenucleus (and of the orbiting electrons) was discovered only by payingattention to the more complex spectra.

In sum, according to Bachelard, the Cartesian approach of beginninngwith the simplest case turned out to be a 'positivism of the first glance'(1934, 154) that emphasized the most apparent features of the hydrogenatom over those that are in fact central to an understanding of it. As itturned out, the spectrum of hydrogen could be properly understood only

64

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gaston Bachelard's philosophy of science

by noting its similarity to complex cases where its essential characteristicswere more clearly present. Bachelard sums up the anti-Cartesian attitudeof modern science toward the simple as follows:

Simple ideas are working hypotheses or concepts, which must undergorevision before they can assume their proper epistemological role.Simple ideas are not the ultimate basis of knowledge; after a completetheory is available, it will be apparent that simple ideas are in factsimplifications of more complex truths (1934, 148).

But if the simple is no more than a revisable starting point, then we can nolonger regard our intuitions of 'simple' objects as the certain foundation ofscientific knowledge.

Regarding the subject that experiences a foundational intuition, Bache-lard notes that to be certain of our intuitive judgment, we must be fullyaware of what the subject itself may be contributing to the judgment. Weneed, for example, to be able to filter out subjective feelings andprejudices that might distort the intuition's presentation of its object's truenature. Now, for such filtering to occur, the intuiting subject must bereflectively transparent to itself; it must be able to see and assess all that itbrings to the intuition. Contrary to Descartes, Bachelard thinks that suchself-reflective transparency is simply not available. The existence ofundetected epistemological obstacles at every stage of scientific thoughtshows that there are always unexamined depths of the self that may bedistorting its intuitions. These unexamined depths likewise undermine thehope of grounding knowledge in intuitive certainty (cf. Tiles, 1984, 39-41).

Bachelard's critique of Descartes can be summed up as rejecting hisproject of attaining clear and distinct perceptions. The requisite claritywould require a degree of self-awareness that is not available, and therequisite distinctness a likewise unavailable simplicity of the intuitedobject.

A fundamental problem for Bachelard's nonCartesian epistemology isthe question of how, lacking Descartes' founding intuitions (or someequivalent epistemic absolute, such as transcendental deduction), he isable to guarantee the objectivity of scientific knowledge. In particular, howcan he avoid the suggestion that what we take for knowledge is just thedeception of as yet undetected epistemological obstacles?

According to Bachelard, the guarantee of objectivity is the movementof the self from the standpoint of its own intuitions and intellectualinclinations to the standpoint of considerations that are cogent for allrational minds. Thus: 'I think that you will think what I have just thought,if I inform you of the rational line of thought [I'evenement de raison] thathas forced me to think as I have' (1949, 58). In such a situation, 'since Irecognize that what I have thought is a norm [normalite] for normalthought, I have the means to force you to think what I think. In fact, you

65

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gary Gutting

will think as I have to the extent that I make you aware of the problem towhich I have found the solution'. In this way, we achieve the 'sanctioning[consecration] of our method, the proof of the efficacy of our thought, the'socialization of truth' (1949, 58).

In this connection, Bachelard speaks of a move from the solitary cogitoto the communal cogitamus (1949, 57) and from individual existence tosocial surexistence (1949, 60). (He also speaks of the community of truth asbased on a corationalism.) His key point is that this move from thepersonal to the interpersonal is a move from the merely psychological tothe genuinely epistemological whereby 'a psychological value becomes anepistemological value' and 'personal knowledge acquires a certain securityby becoming the knowledge of the scientific community [une connaissancede la cite scientifique]' (1949,48). Indeed, he even exploits the psychologicalconnection by developing a notion of an 'intellectual superego' {surmoi)(1949, 75) that is the internal expression of objective epistemologicalnorms. This notion is the basis of Bachelard's complex account of the'intellectual surveillance of the self, whereby rational thinkers monitorand evaluate various levels of their cognitive activity (1949, Chapter 3).

As presented here, Bachelard seems to hold a consensus account ofscientific justification and objectivity. The validity of knowledge-claims isassured by the agreement of the community of inquirers. However, he atone point poses a standard objection to such an account - the possibility ofa false consensus - and, in responding to it, seems to move in the directionof a coherence account. While agreeing that 'two minds can agree in thesame error,' he points out that truth and error are not simply inverses ofone another in the process of the development of inquiry.

. . . error and truth are not symmetrical, as a purely logical and formalphilosophy might lead us to believe. In the sciences, truths group intosystems, whereas errors are lost in a formless mixture. In other words,truths are linked together apodictically while errors accumulateassertorically (1949, 58-9).

But even here, Bachelard may still be giving priority to consensus, since heimmediately goes on to contrast rational systems of truths that are codifiedinto books 'provided with the guarantee of the scientific community' withthe errors found in books that are 'most often characterized by a detestableoriginality' (1949, 59). It seems that, in the end, he does not provide anentirely clear and cogent account of the grounds of scientific objectivity.

Correlative to Bachelard's critique of foundationalism is his rejection ofrealism. The correlation is apparent from the definition of realism he offersin an early book: '. . . realism [is] . . . any doctrine that maintains theorganization of impressions on the level of the impressions themselves,that places the general after the particular, as a simplification of theparticular, that consequently believes in the prolix richness of the

66

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gaston Bachelard's philosophy of science

individual sensation and in the systematic impoverishment of abstractivethought' (1929, 206). Thus understood, realism asserts the primacy andindispensability of the objects of everyday sense experience. It is the view,held by philosophers ranging from phenomenologists to Whiteheadians toordinary language analysts, that the things we see, hear, and touch are theconcrete ('really real') realities for which science supplies merely partialand abstract accounts. When the scientists say that a gas (e.g., the air webreathe) is a collection of molecules, this is only a manner of speaking.Talk of molecules is one way of elucidating some important aspects of thebehavior of air, but there is no question of claiming that air is reallynothing but a collection of molecules, that scientific talk of molecules cantake the ontological place of our everyday talk about air.

Bachelard sees realism in this sense as conflicting with his thesis that thehistory of science advances by a series of epistemological breaks. For therealist, there are no epistemological breaks because every scientificdiscovery is just the addition of another truth about the familiar objectsgiven to ordinary experience. Science never takes us beyond thefundamental conceptual framework that we acquire when we learn ournatural language. Accordingly, Bachelard's first way of criticizing realismis identical with his case for the reality of epistemological breaks. Another,closely related line of criticism is his argument from the guiding role oftheory over observation in the development of scientific knowledge.Bachelard's point here is strikingly illustrated by his reflections on how amodern scientist would approach Descarts' project (in the SecondMediation) of studying the nature of a piece of wax. Descartes, of course,used the wax example against a form of realism that rejects the need forintellectual understanding to interpret the givens of the senses. But this isjust an anti-empiricist account of the nature of our experience of materialobjects; it is not a rejection of the realist thesis (which Descartes accepts)that such objects are as they are given in this experience properlyunderstood.

Bachelard argues that Descartes' approach to the wax is fundamentallydifferent from that of the modern scientist. Whereas Descartes describes apiece of wax as it is in its natural state, 'just taken from the hive', a modernscientist would 'start with . . . chemically pure wax produced by carefulpurification techniques' (1934, 167). Then, rather than observing the waxunder various conditions in which we might find it in our daily experience(rolled between the fingers, softened by the sun), the scientist wouldsubmit it to a series of carefully controlled manipulations designed to put itin a state that will allow the observation of its characteristic features. Thus,he would 'melt this wax in a crucible and resolidify it in a slow, methodicalway. In this way he can obtain a wax "droplet" whose shape and surfacecomposition can be precisely controlled' (1934, 168). Next, he mightproceed to study the surface of this carefully prepared droplet by exposing

67

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gary Gutting

it to x-rays: 'Thanks to the slow cooling of the ball of wax, the surfacemolecules will be oriented in a precise way relative to the surface of thedrop. This orientation will determine the diffraction pattern of the x-raysand yield spectrograms similar to those obtained . . . for crystals.' On thebasis of established results with crystals, the scientist will expect thesespectrograms to provide explanations of many of the wax's surfaceproperties (smoothness, adherence, oiliness, etc.). Bachelard goes on tosuggest how further techniques might be employed to determine themolecular structure of the wax at deeper levels. But the essential pointshould be already clear: scientific observation consists in the systematicmanipulation of an object on the basis of a theoretical preunderstanding ofit. The scientist treats the object as he does because an already acceptedtheory tells him that this is how to reveal the object's secrets. The veryprocess of scientific observation is based on a theoretical redescription ofthe object that characterizes it in terms of very different categories (e.g.,'chemically pure sample', 'orientation of surface molecules') from those ofuntutored experience. 'For science, then, the qualities of reality arefunctions of our rational methods . . . "Objective meditation" [in contrastto Descartes' "subjective meditation" on the wax] in the laboratorycommits us to a path of progressive objectification that gives reality to botha new form of experience and a new form of thought' (1934, 171). In thisprocess, the familiar sensory objects enshrined by realism are left farbehind.

It is apparent that what Bachelard calls 'realism' is a variety of whatcurrent analytic philosophers of science call 'antirealism' and thatBachelard's own view is a form of what they call 'scientific realism'.Moreover, Bachelard's defense of scientific realism has important advan-tages over some more recent defenses. For one thing, it does not presentrealism as merely the result of a peremptory demand for further,theoretical causes of phenomena that are already adequately accounted forby empirical generalizations. Like Wilfrid Sellars, Bachelard bases his casefor realism on the inadequacy of the 'manifest framework' of everydayobservation-language and the corresponding explanatory superiority oftheoretical frameworks. Further, whereas many recent versions of realismfind themselves in tension with historical accounts (a la Kuhn) of radicalchanges in scientific concepts, Bachelard's realism is built on a model ofscientific development that allows such changes. Specifically, his notion ofI'histoire sanctionee provides a way of reconciling the scientific progressrequired by realist accounts with historical discontinuity.

However, while Bachelard provides a valuable approach to the defenseof scientific realism (i.e., of the ontological superiority of the framework oftheoretical science), he is not willing to accept a full-blooded metaphysicalrealism. That is, he is not willing to assert baldly the mind-independentexistence of the entities encountered by science. Instead, he proposes what

68

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gaston Bachelard's philosophy of science

he calls 'applied rationalism' (le rationalisme applique). The term'rationalism' emphasizes, first, the active role of the mind in theconstruction of the concepts needed to describe adequately the objects ofscience. Secondly, it emphasizes Bachelard's claim that, contrary to awidespread empiricist misconception, it is (theoretical) ideas rather thansense experiences that give us objects in their full concreteness: '. . . ideasreveal details and make specific features [specifications] appear. It isthrough ideas that we see the particular in all its richness; thus, they gobeyond sensations, which grasp only the general' (1933, 159). So, forBachelard, theoretical conceptions are not abstractions from the full realityof objects but the way of reaching this reality beyond the vagueness andincompleteness of our sense experience.

We must not, however, think of Bachelard as an idealist who givesabsolute priority to thought. He rejects an idealist constitution of realityfrom the pure thought of a cogito. On the contrary, he holds that, if webegin with mere thought itself (after, for example, a Cartesian 'destructionof the world' through methodic doubt), then any world that thoughtposited would be an entirely arbitrary construction. Idealism must reduceeither to a skepticism that restricts reality to thought or to a 'creationism'that makes the world a mere caprice of thought. Bachelard presents hisversion of rationalism as a middle ground between these two extremes. Wearrive at truth neither by skeptically destroying the world nor by mentallycreating it. Rather, truth results from reason's revision (rectification) of theworld. 'Between the two poles of a destroyed world and a constructedworld, we propose simply to slip in [glisser] a rectified world' (1949, 51).Accordingly, Bachelard's rationalism is applied in the sense that for it anyapplication of concepts by reason must be to an object that has alreadybeen conceptualized: 'rationalism is a philosophy that continues; it isnever truly a philosophy that begins' (1949,54). The objects of our knowledgeare 'already there', not as autonomous givens but as the results of previousapplications of reason.

Bachelard's conception of an applied rationalism is also designed tohighlight the role of scientific instruments in the constitution of scientificreality. The mind's rectification of the world is mediated by the techniquesof experimental manipulation that are an integral part of modern scientifictheories. Indeed, Bachelard speaks of scientific instruments as 'theoriesmaterialized' (1934,13). It is through this instrumental materialization thattheoretical concepts overcome the abstraction of the merely mental andprovide truths more concrete than those of ordinary experience. This ishow Bachelard dissolves the paradox we may feel in his claim that theobjects of theory are more concrete than those of sensation. (Bachelardalludes to this paradox by speaking of the scientific object as an 'abstract-concrete' object.) Science replaces ordinary experiences with its owntheoretically informed experience. It moves us from the 'phenomenology'

69

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gary Gutting

that describes the objects of ordinary experiences to what Bachelard calls a'phenomenotechnics,' which produces objects by the application ofscientific instruments. '. . . science realizes its objects; it never finds themready made. Phenomeno-technics extends phenomenology. A concept hasbecome scientific to the extent that it has become technical, that it isaccompanied by a technique of realization' (1938, 61). Accordingly,Bachelard rejects the classic division that separated a theory from itsapplication and asserts rather 'the necessity of incorporating conditions ofapplication into the very essence of the theory' (1938, 61). It is through thetechnical application of a theory's abstract rational structures that thereappears the concrete scientific object.

Bachelard sees his applied rationalism as a viable ontological middleground between idealism and realism and between empiricism and classicalrationalism. It accepts idealism's emphasis on the mind's active role,regarding any particular object as constructed by the application of theory.But, at the same time, it accepts the realist's emphasis on the transcendenceof the object, agreeing that any particular act of theoretical constructionoperates on an object already given to it. What from one (relativelyidealist) point of view is the mind's rectification of a concept is, fromanother (relatively realist) point of view, a better approximation to reality(cf. 1927, 279). Similarly, applied rationalism tries to do justice to bothclassical rationalism's emphasis on the guiding role of theory andempiricism's emphasis on the need to put theory to the test of experience.'. . . Scientific culture is animated by a subtle [fine] dialectic thatconstantly goes from theory to experience in order to come back fromexperience to the fundamental organization of [theoretical] principles'(1951, 16).

It seems to me, however, that Bachelard's retreat from metaphysicalrealism is less impressive than his defense of scientific realism. His attemptto combine realistic and idealistic views of objects seems to be unstable. Inorder to avoid idealism, Bachelard maintains that any operation of themind is a transformation of an object pregiven to it. But, unless he iswilling to fall back into metaphysical realism, he must agree that anypregiven object itself must be the result of a previous 'constitution' bymental activity. If so, his position reduces to a form of idealism.Consequently, it seems to me that Bachelard's position collapses intoeither idealism or metaphysical realism and does not in fact offer a viablethird alternative.

NOTE

1 Bachelard's work on literature and poetic imagination is fairly well-knownamong English-speaking literary theorists. Mary Tiles' recent book (1984) is a

70

Downloaded By: [York University Libraries] At: 12:42 22 March 2010

Gaston Bachelard's philosophy of science

very interesting and intelligent effort to find inspiration in Bachelard for anapproach to some current problems of analytic philosophy of science. It is notintended, however, as a general guide to Bachelard's philosophy of science in itsown terms.

REFERENCES

Bachelard, G. (1927), Essai sur la connaissance approché, Vrin.Bachelard, G. (1929), La valeur inductive de la relativité, Vrin.Bachelard, G. (1933), L'histoire des sciences dans l'enseignment, Publications de

l'enseignment scientifique, no. 2.Bachelard, G. (1934), Le nouvel esprit scientifique, Presses Universitaires de

France. References will be to the English translation by Arthur Goldhammer,The New Scientific Spirit, Beacon Press, 1984.

Bachelard, G. (1938), La formation de l'esprit scientifique, Vrin.Bachelard, G. (1940), La philosophie du non, Presses Universitaires de France.

References will be to the English translation by G.C. Waterson, ThePhilosophy of No, Orion Press, 1969.

Bachelard, G. (1949), Le rationalisme appliqué, Presses Universitaires de France.Bachelard, G. (1951), L'activité rationalist de la physique contemporaine, Presses

Universitaires de France.Bachelard, G. (1953), Le matérialisme rationnel, Presses Universitaires de France.Canguilhem, G. (1970), Etudes d'histoire et de philosophie des sciences (second

edition), Vrin.Canguilhem, G. (1977), Idéologie et rationalité, Vrin.Foucault, M. (1985), 'La vie: l'expérience et la science', Revue de métaphysique et

de morale, LXX.Smith, Roch C. (1982), Gaston Bachelard, Twayne.Tiles, M. (1984), Bachelard: Science and Objectivity, Cambridge University Press.

71

Downloaded By: [York University Libraries] At: 12:42 22 March 2010