Chapter Three From Theories to Research Traditions · PDF fileFROM THEORtES TO RESEARCH...

FROM THEORtES TO RESEARCH TRADITIONS 71

vies are inevitably involved in the solution of problems; the aim of theorizing is to provide coherent and adequate ions to the empirical problems which stimulate inquiry. ries, moreover, are designed to avoid (or to resolve) the

US conceptual and anomalous problems which their prede-

rs generate. If one looks at inquiry in this way, if one views

ies from this perspective, it becomes clear that the central

itive test of any theory involves assessing its adequacy as a jolt of certain empirical and conceptual problems. Having

loped in earlier chapters a taxonomy for describing the s of problems which confront theories, we must now lay i adequacy conditions for determining when a theory ides an acceptable solution to the problems which con-

Lit. it before we embark on that task, we must clarify what ties are and how they function, for a failure to make some

rudimentary distinctions here has brought grief to more than one major philosophy of science. Entire books have been devoted to the structure of scientific theory; I am attempting nothing that ambitious. Rather, I shall want to insist on only two major points with respect to an analysis of theories.

In the first place, to make explicit what has been implicit all along, C/ic evaluationo ieoLes 1saconz)aretiiYematter What is crucial in any cognitive assessment of a theory is how it fares with respect to its ccmpetitors. Absolute measures of the empirical or conceptual credentials of a theoryale of no sEcarcedecls1vej tastQj tacks / up against its known contenders Mud' of the literature in the phdhdf sience has been 'based upon the assumption that theoretical evaluation occurs in a competitve vacuum. By contrast, I shall be assuming that assessments of theories always involve comparative modalities. We ask is this theory better than that one Is this doctrine the best among the - asaila e op ions"

The second major claim of this chapter is that it is necessary to distinguish, within the class of what are usually called scientific t te-ries between two different sorts of proposi-

tional networks.

In the standard literature on scientific inference, as well as in common scientific pract:ce, the term "theory" refers to (at least) two very types of thing-s. We often use the term theory to denote a very specific set of related doctrines (commonly called "hypotheses' or "axioms" or "principles") which can be utilized for making specific experimental predictions and for giving detailed explanations of natural phenomena Examples of this type of theory would include Maxwell's theory of electromagnetism, the Bohr-Kramers-Slater theory of atomic structure, Einstein's theory of the photoelectric effect, Marx's labor theory of value, Wegener's theory of continental drift, and the Freudian theory of the Oedipal complex.

By contrast, the term "theory" is also used to refer to much more general, much less easily testable, sets of doctrines or assumptions. For instance, one speaks about "the atomic theory," or "the theory of evolution," or "the kinetic theory of H gases In each of these cases, we are referring not to a single theoi butt: a whole spectrum of individual theories. The term

ThreeChapter

From Theories to Research Traditions

The I telitctualjwecOon Of

on esttthliehed conceptual scheme is to dcternnne the

patterns of theory the tneam

ingful questionA the legitimate

inzernreta!ions S T000MIN (190), p. 40

lutionary theory" for instance, does not refer to any single

y but to an entire family of doctrines, historically anti

udv retated, a:i of v,11 I, work from the assumption

rganic species have common lines of descent. Similarly. term 'atomic theory' generally refers to a large set of tines, ak of which are predicated on the assumption that er is discontinuous. A particularly vivid instance of one ct which Inc l udes a wide variety of specific instantiations is cJ by recent "guantum theory Since 1930, that term has ioen among otner thmgs) quantum field theories, group nes. so-cdlled Smatrix theories, and renormalized field r:es—between any two oi which there are huge conceptual

Ic differences between the two types of theories outlined

e are vast: not only are there contrasts of generality and

ificit bets yen them. but the modes of appraisal and

atin app:'priate to each are radicahy different. It will be

tentra clai:ts of this chapter that until we become mindful -

t cognitive and evcivatinal di/ferences between these two

011 eherjncs, it wi/i he an posihIe to have a chewy of !

utica arugress WZL It is hit torccailv sound or phdosoplitccillj' / tame, 4 et it is not on l y fidelity to scientific practice and usage p.

M requires, us to take these larger theoretical units stiou1y.

Pt J the research done by historians and philosophers of

me :n the last decade suggests that these more pd,nits -

inanuis exhibit many of the epistemic features which,

ch riidth ch rhTKic B'Theanaystwho -.

Is his range to theories the narrower sense. Specifically,

as been suggested by Kuhn and Lakatos tha t ,the more

'rd iheoriey, rather than the more specific ones, are the 't -

tar

4-01JI eor understanding and apJ/raiscgL,Jentlfic -/ eK.

share this conviction in principle, but find that the accounts

erto given of what these larger theories are, and how they

VC. are not fully satisfactory. Because the bulk of this

[)ter will', be devoted to outlining a new account of the more

)al theories (which I shall be calling research traditwus), it is

i'opriate that I should indicate what I find chiefly wanting in

the best known efforts to grapple with this problem. Of the many theories of scientific evolutior that have been developed, two specifically address themselves to the question of the nature Of these more general theories.

Kuhn's Theory of Scientific "Paradigms"

In his influential Structure of' Scientific Revolutions, Thomas Kuhn offers a f scientific progress whose primary element is the ." Althc-uglt Kuhn's notion of paradigms has been shown to be systematically ambigppus (and thus difficult to characterize accurately), they do have certain iniiEa1jactjstics They are, to begin with, "ways of looking at the world" broad quasi-metaphysical insights or hunches about how the phenomena in some domain should be explained. Included under the umbreja of any well-developed paradigm will be a number of specific theories, each of which Presupposes one or more elements of the paradigm. Once a paradigm is accepted by scientists (and one of Kihn's more eëj'a1I1sIstlat in any "mature" science,2 evec scientist will accept the same paradigm most of the time), they can proceed with the process of " EEiiam articulation" also known as "normal science." In periods of normal science, the dominant Paradigm wiil itself be regarded as unalterable and immune from criticism. IidividuaI, specific theories (which represent efforts "to articulate the paradigm," i.e., to apply it to an ever wider range of cases) may well be criticized, falsified and abandoned; but the pa:-adigin itself is unchal-

L_~indjcaV~s ONV tTT's -point is determined) that

scientists begin to ask whether the dominantparadini is rea a period of "crisis." DOTi'is scientists begin for tire first time to consider seriously alternative paradigms. If one of those alternatives proves to be more empirically successful than the former paradigm, a scientific revolution occurs, a new paradigm is

recognizes clearly that maxi-theories. different co(;,nitive

There is much that is valuable in Kuhn's approach.,

FROM THEORIES TO RESRARCH TRADITIONS

;4 lriytief unctions than, mini-theories. He has probably e the fievr -thir to stress the tenacity and perseverin,,

S of clobl thPories—even when confronted with serious ik"JomJies He has —eart'ectivreiected the (widely assumed) *dative character of science._' But for all its many strengths, nhns mod-.i of scientific progress suffers from some acute

and empirical difficulties. For instance, Kuhn's iinl Th; m.radignis and their careers has been extensIvely

bcizcd by Shapere, who has highlighted the obscure and

uustsencies n pnsusQLthaonon. Feyerabend and hers have stressed she historical incorrectness of Kuhn's

is nan' ' atttpica or ),rn0aL V1r5U0VeVerVmaOrjeriOdW the history QfaC1eI1Ce crocterized both by the co-existence of numerous com-

Lf —n fi fting jSZBcI u by the persistent and continuous manner in which the

datlda e scientific community. Numerous critics have noted the

Kneory of si if (as Kuhn says) a few do not produce a crisis, but "many" do, how does

t scientist determine the "crisis point?" There are other tiovs flaws as well. in my view, the most significant of these

1. Kuhn's failure to see the role of conceotual problems in scientific debate and in paradigm evaluation. Insofar as Kuhn tants that there are any rational criteria for paradigm choice,

or for assessing the "progressiveness" of a paradigm, those criter:a are the traditional positivist ones such as: Does the theory explain more facts than its predecessor? Can it solve some empirical anomalies exhibited by its predecessor? The whole notion of conceptual problems and their connection with progress finds no serious exemplification in Kuhn's analysis.

2. Kuhn never really resolves the crucial question of the u /ats/upb eenapaiacIigmantscon.swuenttheoi 'cc ion Does the paradigm entail or merely inspire its constituent theories? Do these theories, once developed, justify the paradigm, or does the paradigm justify them? it is not even clear,

FRO-M THEORIES TO RESEARCH TRADITIONS 75 in Kuhn's case, whether a paradigm precedes its theories or arises no/ens S'Olefls after their formulation Although this issue is exr-

enel) complex, any adequate theory of science is going to have to come to grips with it more fficctt1v than Kuhn has. 3. Kuhn's paradigms have arigidi of structure which pmcluesfenifroj evolving throtg1r the CotF1c of time in response to the weaknesses and anomalies which they generate. Vloreover, because ne makes the the

there can he no corrective l'efatzonshtp between the Accordingly, it

the inflexibility of Kuhnian paradigms with the historical fact that many maxi-theories have evolved through time.

.

4. Kuhn's "disciplinary inatrices," are always As a result, Itisdifticult to

understtd how he can account for the theoretical controversies which have occurred in the developrneit of

r_eun_i,ZTy only debate aboutff assumptions which have been made When, for instance, aKuljan maintains that the ontological and methodological frameworks for Cartesian or Newtonian physics, for Darwinian biology, or for behavioristic psychology were only implicit and never received overt formulation, he is running squarely in the face of the historical fact that the core assumptions of all these paradigms were explicit even from their inception.

ideD lied by Pointing to their "exenji I

5. Because paradigms are so infliDlicit and be (basically an

archetypal application of a mathematical formulation to an experimental problem), it follows that whenever two scientists utlhze the sametars t are for Kuhn, ipso facto

i!T_TiTf'~rent scientists often utilize the same laws or exemplars, yet subscribe to radically Giver ent views about t e most basic questions of scientific Ontology and

thoQlo.y (For instance, both mechanists and energeticjsts accepted identical conservationlaws.) To this extent, analysing science in tscience ms of paradigms is unlikely .to reveal that "strong

FROM THEORIES TO RESEARCH TRADITIONS 77 But against that, Lakatos' model of research programmes shares many of the flaws of Kuhn's paradigms, and introduces some new ones as well: 1. As with

is exclusively enlplrlcal;the only progressive modifications in a theory

those which increase the scope of its empirical claims. 2. The sorts of changes which Lakatos allows within the

e0ries '~v_5_`lcc~onstifu7tE his research programme are

as the relation between any theory and its successor within a research programme, the addition of a new assumption or a semantic - .

re-interprJat1onoffl1slfltltepledectl On this rrabe view of things two theories can only be in the same

if one ,of the two entails the / er s we Shall see shortly, in the vast majority of cases, the succession .. .- --- ----------------' .

Of specific theories within a maxi-theory involves the elwzlnatwn as well as the audition of ass,tioiis and there are rarely Successive theorieswhich entail U

J. A fatal flaw in the Lakatosjan notion of research programmes s its dependence upon the T iPop notions of cHi-~-ir-i-ca-f--al-i-a--To-2,caI content."

All Lakatos' measures of a comparison of the emJLniiLery

member of the series of theories which constitutes any research I -

ingly, the attempt to speci content measures for scientific theories is extremely problematic if not literal l hI 15/

Parisons of cont2aL are ei ~_= his followers have been able to identify

historical cas any

progress can besI1vnsjctlity ii 4 Because of Laao idiosyncratic atic view that the acce Lance of theories can carce 'if ever be rational, he cannot translate

his asscssnents of progress (assuming he cikb{ff1!) icommatjonsa out cognitive action.' Although one research programme may be more progressive than another,

/ we can on Lakatos Recount deduce nothing from that about -

I

FROM THEORIES TO RESEARCH TRADITIONS

:ork of commitments—conceptual theoretical, instrumen-and metaphvsical' ' which Kuhn hoped to localize with his ry of paradigms.

Lakatos' Theory of Research Programmes'

.otrel in response to Kuhn's assault on some of the of SflCL, 1mm

an a.::..ive tnr:, about the role of

, se "super-theoT-ics- in the evolution of science. Calling such

,,,era! -:aatmes. Laka tos arucs that

programmes have three ---elements: I a hard-core" ne~-rative 1111,2~urisuc") of fundamental Jart:ptMns '-:::LO

'inoi abandoned or modified without rcpuuJ_iation of the proutramm 'msn hcurlSu, '.1 O1CO

-1tains "a partially articulated set of suggestions or 'tints on • - specific - - a mon:. sophisticate sw i our

les whenever we wish to imnru'.e item. and (3) "a series

- DSCI *

-sults from adding au.:: - clauses ... the plCViouS

- H. - Such theories are tne cit intar:iations of thc - research :irrann:_. Research programmes can be

vorietv of ways; but progress, for

S :m- '. more than for Kuhn, :5 c IDECLOn exclusively of - empirical of aumur. it is the PDSSCSS100 of

empirical. content," or of a h igher"degree 01 empirical - -

which makes one theory superior to, and more

Lakatos' model is, in many respects, a decided improvement Kuhn's. Unlike Kuhn, Lakatos allows for, and stresses, the

storical importance of the co-existence of several alternative search programmes at the same time, within the same mair lfloKuhn, who often takes the view that Rarad ig ms 0 in gnimensurible13 and thus not open to rational compari n, Lakatos insists that we can objectively compare the relative ogress o c

pet, i research traditions. More than Kuhn,

tfries to grapple with the thorny question of the relation the super-theory to its constituent mini-theories,


and a theory of rational acceptability (or, to use guageeTriethodological "appraisal" and

I

Lakatos' claim that the accumulation ofanomalies has no hoUrfleOfl the am I

I ttIad y the storv of science. akat,os' research njro~~, s, like Kul's paradigms,

tc rgid in their hard-core structure and admit of no funda- . menta chan;es. I

VEtshod be clear, even from this very brief survey of two 4 the major theories of scientific change, is that there are a

ofna1v'danonistoricaldiff1culiesconfroi tin,-

•stti attempts tc understand the nature and role of maxi- Iheories. \Vith some of those diffidilties in mind, we can turn

explore an alternative model of scientific progress, built

/ UI an dements outlined in the previous chapters. One crucial (ott of that m rdcl v."111 be whether it can avoid sonic of the

.

t gobcms vhch handicap its predecessors. Although there are putterous common eiements between my model and those of Kuhn and Ldcatos and I readily concede a great debt to their ;ioriecring svorkh them are a sufficiently large number of differences that I shah try to develop the notion of a research

ttddi;ien mru or !et Iron: sratch.

FROM THEORIES TO REsEARCI.1 TRADITIONS 79 2. L—LeLy research tradition exhibits certain metaphysical and

individu- ate the research 3. Each research tradjtjoii (unlike a specific theorygo

through a number of different, detailed (and often mutually contradictory) formulations and generally has a long history

theories are frequently -~_Inificant period of time. (By contrast,

These are by no means the only important characteristics of research traditions, but they should serve, for the time being, to identify the kinds of objects whose properties I would like to explore.

In brief, a research tradition provides a set of guidelines for Part of those guidelines

constitute the

specifies,

types of fundamental entities which exist in the domain or domains wJthinwlIch the research trad ition is embedded The

is to y"reducing"

them to the ontology of the research tradition. If the research tradition is behaviorism, for Instance, it tells us that the only legitimate entities which behavioristic theories can postulate are directly and publicly obscryabi physical g1ld physiological signs. If the research tradition is that of Cartesian physics, it specifies that only matter and minds exist, and that theories which talk of other types of substances (or of "mixed" mind and matter) are unacceptable Moreover, the research tradition Outlines the d~ferent nzodes bj, which these entities cat, interact. I nus, Cartesian particles can only interact by contact, not by action-atadjstaice Entities, within a Marxist research tradition, can only interact by virtue of the economic forces influencing them.

Very often, the research tradit rtain rnodes of procedure -which constitute the legitiniate."iet/ inquiry

open to a researcher wi!bip_that._LLaAit4,on Of

.

These be wide-ranging in scope, addres-

sing themselves to experimenta1technig

For instance, the methodological Posture of the scientist in a strict Newtonian

—T.. /

: FROM THEORIES 'C 0 RESEARCH TRADITIONS 81 allowed 1-yell to explain the geological past. Similarly, the mathematical ontology of the Cartesian research tradition (an ontology which argued that all physical changes were entirely changes of quantity) was very closely connected with the (matliematicail' inspired) deductivist and axiomatic methodology of Cartesianism, As we shall see later it does not always happen that tile oo1oafldmcthd! tradition arclosely intertwined(for instance, the inductivist methodology of the Newtonian research tradition had only the weakest of connections with that tradition's ontology), but such cases are the exception rather than the rule.

So a preliminary, working definition of a research tradition • could be put as follows: a research tradition j,g,,jeral

\ • aJ7l 3tions about the entities andprocesses in a doinain of p. study, and about the appropriate methods to be used for —R~~~ins and constructing the theor*—' that domain.

Theories and Research Traditions

Every research tradition will be associated with aseries of specific theories, each of which is designed to particularize the ontology of the research trMclition and to illustrate, or satisfy, its methodology. The mechanistic research tradition in seventeenth century optics, for example, includes several of Descartes' theories as well as the optical theories of Hooke, Rohault, Hobbes, Regis, and Huygens.20 The phlogistic tradition in eighteenth century chemistry received more than a dozen specific theoretical formulatjons, 21 Man), of the theories within

V inconsistent rivals, precisely because some theories represent attempts, within

their predecessors.

The individual theories constituting the tradition will generally irically testa e they will entail (in conjunction with other specific theories) some precise predictions about how objects in the domain By contrast, research 1~', traditions are ,zdrifrea, /ii

a

dO TS I

FROM T000RIES TO RESEARCH TRADJTLONS

h tradition is inevitably inductivist, allowing for the d of only those theories which have been "inductively

from the data. The methods of procedure outlined behavioristic psychologist are what is usually called donailat." Put sinsplisucailv. a research tradition is thus

icil and ,ne:iwd jet! '10 's 'and "don'ts.

•ii attempt v bat is forbiddon ha the metaphysics and metho-p1ogv of a research tradition is to put oneself outside that fttshtion and to repudiate it. If, for instance, a Cartesian pliscist starts talking aboot forces acting-at-a-distance. if a behaviorist starts talking about subconscious drives, if a Marxist begirts speculating about ideas which do not arise in e;'cnse to sko economic substructure; in each of these cases, Iha activtrv indicated puts the scientist in question beyond the pale. By breaking with the ontology or the methodology of the tss:arcb tradition within a hich he has worked, he has violated dse strictures of that research tradition and divorced himself rwnt it. Needless to say, that is not necessarily a bad thing.

taii, :evolutions in scientific thought havC come from thinkers who had the ingenuity toJggjkwith

a rose c d cm m a and to ruurajborI

But ' hat ac mrst pt -r if w are to understand either the

jogiC or the history of the natural scicnces is the notion of the

H a reseach tm tion for it is precisely that integrity nhich stimulates, defines and delimits what can count as a tolutioa to many of the most important scientific problems. 19

Although it is vital to distinguish between the optoipgIcal and the methodological components of a research tradition, the two are often intimately related, and for a very natural reason: ndiTieiv, that one's views about the appropriate methods of inB'dfT'PBrbillf dO ifpkf dt iBPsBUout the

Charles Lyell defined the "unlforn1itarian" research tradition in geology, his ontology was restricted to presently acting causes and his methodology insisted that we should ''explain past effects in terms of presently acting causes." Without a "presentist" ontology, his uniformitarian methodology would have been inappropriate; and without the latter, the presentist ontology would not have

FRO THEORIES TO RESEARCH TRADITIONS FROMTHEORIES TO RESEARCH TRADITIONS

4ib!e. The rve' v gciieraitv. as well as their normative 83 is (or ought not be) to judge that tradition false.

1frmt ts rejectinas momentarily 0 t;L na lOCOSSO

c level

unsuccessful are we necessar ilv relegating It to pe1rnaneit oblivion; to the 5(u) a the abstract is contra e can exp11c11pulatecolldjtIois 44 (if, a' ShOaIdCSJd studied ieea cdos do which if satisfied wouH revive and ecussitate if o ddanswers to specific qSQS A research it Thus when we reject

a research tradition ( yp g

e are merely making tp

atiye decision not to utilize It for the frctad all, an inte4face beveen water and air; it will not tell moment be there is an 1 te rative to at which has proven to be a

: tshat happens if we put an eight-month-old female mouse . • more successful problem solver. Just as the fortunes of

o a maze; it will not tell as why lead melts at a lower a research tradition are linked closely to the problem-solving effectiveness

UI erature thai; copper. But it would be a mistake to conclude of its constituent theories, so too is the determination of the •m the fact that research traditions do not offer solutions to adequacv of theory bound U') fie U er at they outside O the problem-solving

with an assessent of the efll5fleffectienes m

of the T : f research entire set of theories S awned b l

theory On Is \ needfo (As

is a e shall 'tradition, henwatever the problem -solving merits late tae-eaareb tradition even goes so far as to define of that

particulat theoryit is likely to be regarded ! ill i problems are, and what importance soLi1d be r is highly suspect

For instance COUOL Rt mfoid s theories hcit 'Itd of conduction and convection were Ir a i ; ,.1nTT-T:r the

superior to any alternative theories of thermal flow in fluids available shkn 19 like a jjr;

in the petiod fron 1800 o 1815 Nonetheless fe scientists took Rumford 's theories tradionhich makes no predictions wi ch solves J kVecific problems, which is fundamental y no, mat v and ; . •-

seriously because (as they saw it) the research tradition in which Rumford worked ts3ial—aauld be objectively evaluated may seem pa - (deriving from Boerhaayc) had been discred

ted by the emergence Ktaut nothing could be further from the case, for we can . . of rival research traditions in chemistry

&especially Joseph Black's), quiresimply that a sccessfiii researchtraditionis one which suggested that heat was a

substance rather than, as Rumford tih 1J conivonenttIieoricstotheadequatd

imagined, the random mot ot of pt'tic1cs Rumford 's specific theories _Ion ii increasing j1zeofenpuca'andcoz uJ only became - t-Ishionable in, t3 e 1840s and 1850s because ining whether s tradition is successful in this _ by that time the nce between %dljOus does not mean, of course, that the tradition has been , research traditions had shifted sufic ienfly that many scientists rtitfirrned'

or refuted. " Nor can such an appraisal tell us were more prepared to consider seriously specific theories (like Rumford's)

ything about the truth pjifa1sity of tradition.22 A research which grew out of a kinetic research tradition. •

1JtQn_ind be enoimousiy Contrariwise, a theory even , 4t(flics and et fIa\edTh 1s-ontolog) or methodolog Equ sll an inadequate. one will have some strong arguments in favor 01 can conceive that a research tiadition iiight be true, and

its if it is hal ed v ith a research t aditton that is otherwise highly

( (perhaps because of the unimaginativencss of its propon- successful Thus, theories of mechanistic physiology in the late . its) unsuccessful at generating theories which were effective

seventeenth century (such as

those of Borelii and Pitcairn) were highly p I)blem solvers nce to abandon or reject a rcseach tradition regarded in many

crcles where the mechanistic lesesecli tr'iditjii was flour h ,'

lslnb

FRO,I THEORIES TO RESEARCH TRADITIONS 85 Similar considerations

ap lY to the rcvcrsc relation between theories

777 and research 'traditions. For examptc, given tile theory

oflaetasde\.e101)ecibyHtlygeIIs we cannot deduce the basic assumptions of the research tradition within which Huygens worked (We may, of course, be able to deduce that Huygens was working in a research tradition in which collision phenomena Constituted an impor

tant unsolved problem, for if not, why should Huygens have bothered working out a theory of colli-sion?). But it is never possible to deduce the whole of a research tradition from one, or

even all, of the theories allied to it. The reason why entailment will not help here

is very simple: there are a number o mutually lncoflsjstemit theories which ca n ch4LflU lance to the sonic research tra number of dt'jfe,'emzt research traditions which Ciif

Täe Pro the s osition1j/ base for at a1pies of both phenomena abound: many scientists in the Cartesian optical tradition argued that light travelled faster in optically denser media; other theorists, within the same tradition, asserted the converse Staying within the history of optics,

there are numerous examples of competing research traditions claiming to justify the same theory. For instance, Newton's theory that light has

was accepted - alike by scicntists in v. 00

~-,]d car USCU!flt ti' entailment were the relation between research traditions and

theories, then it would be impossible for such shuations to arise. Since the relation we are trying

to explore is evidently not one of entailment, what can we say positively about it?

wo nd research traditions are related: one is ijis,roricell 5

cOca t e other is we c it matter it orical fact that 11110st if not all of the major theories of science have emerged when the scientist who

invented them was working within one or another specific research tradition. Boyle's theory of developed gases within the framework of the mechanical philosophy. Buffon's embryo- logical theories were developed as efforts to apply the Newtonian research tradition to biological phenomena Hartley's theories

of sensation were developed within the research tradition of associati000jst psychology. Hertz's electrical theories were linked in important ways with the Maxwelljaii 1-,-search Tradition.

,01

fe


a though, hrdgod entirely, on their own merits, they were goficantiv inferior to certain theories in other, less successful

arch traditionsk 4 I4 to this point, I have been deliberately vague about ssilbing the dnd of relation which exists between a theory ti its"parent" research tradition. I have spoken of research d stiens inspiring' or "containing" or "generating' theories, o about theories 'oresupposing" or "constituting" or even

nin" research traditions. This is an extremely complex ll'e:: the ambiguiti; of the meta hors 1 have invoked to atacterize the theory research tradition connection is a optom of the difficulty of tackling this problem head-on. Hot that tusk cannot be further postponed. I shall begin by Ving hat the relation beuveen theories and research tradi-

is arch

oi.tions Co riot entail their component theories; nor do those 1BbTg, taerrthtt FjoiTf entail their parent

gnt then

s ould be a simple matter to determine mechanically which tories belonged to any given research tradition, or the carolc eraditionis lurking behind any theory. But to see the tory research tradition connection in such formal terms is 1lpitfoiv to niisurtderstand the differences in kind between

A reoearch tradition, at best, specifies a general lologv foature, anOageneiT met od for solving natural

wirimga given natural ther

n urtrcuatesaverv speci ic an anunioer 01 nature. Tobe told, as the

vtBiTtan—research tradit;on in mechanics tells us, that we aId treat all nonrect1lnear motions as cases of centrally ected forces, does not entail any specific theory about how to )lain, say, the motion of a compass needle in the vicinity of a7. 'rent-carrying wire. To develop a "Newtonian" theory for that Ao tticular phenomenon, we must (as Ampere did) go far beyond deductive consequences of the Newtonian research tradition. be told, as the nineteenth-century "mechanical" research

dition tells us, that heat is simply a form of motion, does not/ luctively lead us to Boltzmann's version of the kinetfd ory of gases or to statistical thermodynamics

FROM THEOJES TO RESEARCH TRADITIONS

A sofc dhcory, abstracted from its historical context, may not gie unambiguous does as to the research tradition (or tradiions) with which it is associated. It is just this fact which has lead :tlaa: scensts and philosophers to imagine that

are ustuih appraised and assessed independently of the research traditi cns of wh:cn they are a part. But we should not be mi lcry fact that a theoa taken abstractly, does not havc its "parent" research tradition stamped all over it. Histc:iea research can always 'a: least in principle) identity the re earc' tradhb n(s) with which a particular theory has been aciate,in :hssense, the connection between a theory and a raraditin is as real as any fact of the past, and it is as

as the most important facts of the pas.\ order to sea 05" imoortant these connections are, we need to look at the

ays m wn;cn :naories anO research traditions can Interact. TEtt"rtncst iOl5fffI:tnsotesoiInteractioiT'Ereoenerallv

mtuen ces at toe research tradition upon ltz theories. These influencas take a variety of forms:

The ;7roblem determinin.,&roleW research tra 16 Even b continue uso thereafter, a research tradition will often strongly influence though it does not biIy determine) the _range and the shting of the empirical problems with whiEiits component

have a deci'' nf nce on what can count as the range of possible concapt:ta. nwcmems whica the theories in that tradition can generate. Tne'c two processes are important ones and should be discussed in se me detail.

i. Among the other roles of a research tradition, it is designed to delimit, at least partially and in outline, the domain f)J alpJicaii of its constituent theories. It does this by indicating that if is appropriate to discuss certain classes of empirical problems in the given domain, whilst others belong to foreign domains, or are "pseudo-problems" which can be legitimately ignored. Either the ontology or the methodology of the research tradition can influence what are to count as legitimate problems for its constituent theories, if, for instance, the- Methodology a research tradition ypgjfisJ1 usually

FROM THEOI?,iEc TO RESEARCH TRADITIONS 87 ill—eertain experimental tegfpue which alone are the

legitimate investigational modes for determining what are the data to be explained, then it is clear that only "phenomena" which can be explored b" those means can, in principle, count as legitimate empirical problems for theories within that tradit,o:'. A classic example of this process is offered by nineteenth-cc ntLi-r y phenomenological chemistry. Scientists in this tradition argued that the oily legitimate problems to be solved by the chemist were those which concerned the observable reactions of chemcial reagents. Thus, to ask how this acid and this base react to form this salt is to pose an authentic problem. But to ask how atoms combine to form diatomic molecules cannot conceivably count as an empirical problem because the niethodo:agy of the researe:i tradition denies the possibility of empirical knowledge of entities the size of atoms and molecules. For other research traditions in niI)eteenth-centui.y chemistry, questions about the combining properties of certain entities not directly observable constituted authentic problems for empirical research, 25 (Contemporary behavioristic psychology and quan-tum mechanics likewise have methodologies which strongly preeluce from consideration as problems certain "phenomena" which other research traditions countenance)

tradition may exclude certain sltuaticns from, or include domain, Thus.- tile rise of tile Cartesian mechanistic research

seventeenth century radically transformed the accepted prohem domain for optical theories. It did so by arguing, or rather by simply postulating, that problems of perception and vision'—prohlens which had classically been regardej as legitimate empirical problems for any optical

theory—sLou be relegated to psychology and to physiology, fields outside the domain cf optics, so that such empirical problems coulo be safely ignored by the mechanistic optical tOeorjst.

A different kind of example is provided by late nineteenth-century physics, where the sibtle fluid tradition (of Faraday, Maxwell, Eertz, and others) countenanced as legitimate empir-ical problems clueries about the properties of the electromag-netic aether. Indeed, the classic Michelson-Morley experiments

EEC CI THEORIES TO RESEARCII TRADITIONS

riginaid: conducted in order to determine the drag fncent Of bodies moving through such an aether. With the

teritonce of sgec;al relativnv theory. however, a new research ,diti a and its related on:cicgv cut out from the domain of

encairicto prebtem of hvsics all questions about the ,sticitv, Ia:, and velocity of the nether—questions which

been ceritral empirical problems between 180 and 1900. wse few examples shctdd make it clear that research lemons can pin': a dcci' lye role in specifying, the sorts of ;ags that are to count as potentially solvable empirical

biems for dm1: constitucat theories. Equaib' important is the v. ay In which a It tradition

I', oenerat conceptual problems for i t s constituent theories.

FROM THEORIES TO RESEARCH TRADITIONS 89 The constrajiuto role of 'search traditions. As we have

already said, it is the primary function of a research tradition to establish a general ontology and methodology for tackling all the problems of a given domain, or set of domains. As such, it acts r-.cgati-,,,cjy as a coistra

1 * '"t oil the types Of theories which can be developed within the domain. if the Ontology of the research tradition denies the existence of forces acting-at-a-distance, then it clearly rules out as unacceptable any specific theory which relies on noncontact action, it was precisely for this reason that "Cartesians" such as Huygens and Leibniz (committed to an ontology of pushes and pulls) found Newton's theory of celestial mechanics so otiose. Einstein's theory of the equivalence of matter and energy excludes from consideration any specific theory which postulates the absolute conservation of mass. The mechanistic tradition in heat theory (with its corollary that heat can be turned into work) precludes the development of theories which assume the materiality of heat, or heat conservation

There are also many occasions where the methodology, of a research tradition rules out certain sorts of theories For instance, any research tradition which has a strongly inductivist or observatjonahist methodology will regard as inadmissible "specific" theories postulating entities which cannot be ob-served. Much of the opposition to subtle fluid theories in the eighteenth century and to atomic theories in the nineteenth century was due to the fact that the dominant methodology of the period denied the epistemic and scientific well-foundedness of theories which dealt with "unobservable entities."28

III all these cases, the research tradition within which a scientist works nreeludes him from adopting specific theories which are incompatible with the metaphysics or methodology Of the tradition.

Thus far, we have focussed attention primarily on the In which research trad'tioll",-x-c-IkLd-L--certain

also have,howeve Positive function

r, two very s.

The heuristic role of research traditions. Precisely because they postulate certain types of entities and certain methods for

FROM THEORIES TO RESEARCH TRADITIONS 91 temperature gradient between input and output corresponding to the top and bottom heights of the waterfall. It is in terms of this analogy that Carnot develops the "proof' of his theory. It is clear that, if Carnot had not conceiycd of heat as a conserved substance capable of flowing from one point to another without loss of its quantity, lie almost certainly would not have enunciated his theory. But that way of conceiving heat was a natural result of the research tradition within which Carnot worked

One final example may make the point still clearer. When Descartes attempted to develop a theory of light and colors, he had already defined his general research tradition. In brief, it amounted to the assert

ioni that the cnly properties which bodies can have are those of size, shape, position, and motion. The research trad i tion did not, indeed could not, specify precisely what sizes, shapes, Positions, and motions particular bodies could exhibit. But it did make it clear that any specific physical theory, in optics or elsewhere, would have to deal exclusively with these four parameters As a result, Descartes knew—when he set out to explain optical refraction, the colors of the rainbow, and the path of light through lens and prismsthat his optical theories would have to be constructed along such

lines. So he sought to explain colors in terms of the shape and rotational velocity of certain particles; he explained refraction in terms of differential velocities of such particles in different media. Moreover, since his general research tradition made it clear that particles of light are exactly like other material bodies, lie recognizeti that he could apply general mechanical theorems (such as the laws of impact and the principle of conservation of motion) to a theoretical analysis of light. Again, none of his theories followed logically from his research traditioni' but, in the ways indicated, that research tradition directed the construction of Cartesian theories in a number of subtle and important ways.

In all the cases mentioned thus fa:', the research tradition functiojis leuris ica y o suggest an

initial theory for some JO ant euristic role for tile research

tradition as Lakatos has pointed out, arises when one of its

t1ttnent theories requires modificationiuse of its Ink f

FROM TREORTES TO RESEARCH TRADITIONS

eFt1atog : propries ot those entities. research tradiiolls T :ay a ii heuristic role in the construction of specific

eiitific heors. Not of course because theories can in any wose be deduced from research traditions; but rather because

MZIC.t )h can urov:de vita! clues for theory construc-p. Cetes:act- mc case at BJTMin Franklin and his efforts to I eu Into a tlhwor%- of static eiectric:tv. Franklin was fa m1 tar

g no man tprtculartv, eleetrficaton by friction, 'kmraecones, and the Levden jar). Working Witfllfl a research idEticn which postulated the existence of electrical matter. rarkhn -needed a Oscars which could explain how, friction

'lectrifies bodies. how electrical bodies could attract and repel, it ectdd be tered in a condensor, and why certain

wore LOfl actors and others were insulators. In the 'am stages of the development of his theory, Franklin came to he view ibm nositive electriPeation consisted in the accumula-i ta "Ethin bodies, of an excess amount of this electrical fluid, ittbe negative electrification was caused by a deficiency of this hid. f tbes specific theoretical assumptions are linked ogedser with rise ontology of his research tradition, an ontology dstch rostalared that electricity -was a form of matter and borefore conserved in the same way that ordinary matter was, I became natural to assume that electrical charge must be misers ed. This important theoretical insight, subsequently onfirmed in Franklin's experiments, emerged as an almost aevisab:e result of Franklin's thinking about the relations etsveeis his emerging theory and its parent research tradition. did not foh'v logically from either the early theory itself, nor

the research tradition. It was the juxtaposition of the two .wt made possible this vital theoretical extension. A differ-,nit sort offetii'is'iic e is illustrated by the early

istorv of thermodynamics, When Sadi Carnot set out to eelop a theory of steam engines, he sought to do so within the tsearch tradition of the caloric doctrine of heat. Within this 'adition, heat was conceived as a material, conserved substance ;ipable of moving between the constituent parts of macroscopic odies. Carnot, familiar with the work that could he performed y such simple mechanical systems as a water wheel, tried to )nceive of heat flow on analogy with the fall of water, with the

92 FROM THEORIES TO RESEARCH TRADITIONS

aiitv1. An

sound research tradition will be

Ic iv iniprove its J)roblenl -SoIvInI,J

FROFI THEORIES TO RESEARCH TRADITIONS 93 phenomena as elasticity and gaseous mixing. Had Hales been working in research traditions other than the Newtonian one, such an assumption would have been unthinkable or at least in fler of

••If Fcr I nstance, when earvversions of the kinetic theory of •;

---- JusIIrlcatjon At a minimum, his theory

would have had to address itself to justifying :w a aroato nne serious predictive failures, there that assumption) But, as a Newtoni-111, ,th in the research tradition which

1_Il1I Hales cou d assume, almost without argument, that it

t d Li r5 rel rodnfwitoiistItm gIll u r was appropriate nd leCitjnate to conceive of ses s

;deio cAgree fid6nvere needed to accommodate - - swarms j j of n tully ILpCJfCflt ticks By sanctioning certann

cI losses. kimtcists could introduct molecular -

assumptions in advance, the research tradition thus frees the scientist working ssn Or alter thair assumptions about molecular elasticities. If within it from having to justify all of his assumptions,

'es hA oat condense in accordance with theoretical predic- and gives him the time to pursue specific problems

:h addition of weak kstermolecuiar attractions could do of interest outsj th h

4: '-

J ab. All these, an m n d ay similar "gambits" emerge quite cientist lor constructing theories based on

-I assuriptio psA:v fmm ragarding matter as possessing a molecular ~ks that his )riMUMI /Y —fellow researchers

i L . c r within the sam dt hic find Jut

-.- AA '

t' Research

within them three classes of assumptions:those which are

ecause by those wh ich L are .forbidden bv the research of course, those i\\hjle 'lot forbid db

research tradition Itself pl-ovide's ULU the theory (for the

CIC1111sts wor 1119 within ally one research tradition, there will be a broad COflSCOSUS about where any given statement falls as between those three pigeonholes.

Swmin the discussion thus fthat such ions can ' L*f Seen

their theories make they can serve to stamp certain theories as

CY Call influence the of edipiricai and--ceptU problems forth

e ir ieories and

the

The Separability of Theories from Research Traditions

Up to now, I have stressed that virtually all theoretical activity takes place within the context of a research traditi, that such traditions con on

strain, inspire, and serve to justify the

F


:;;co:es vhici are subsume under them. Without wishing to

Hoa:o of that, it is equally important to recognizc that

:i'e:s are ci:oum stances ; m

the research ire, pired or justified

Ga1iso's thco:y : f fai. f istance. has (since the 1650s) been

seoaratov fooin the Galilcan rescarob tradition; similar oro1i s said abit Pasteur's theory of disease,

's :hcor of eceti-omagnetism. Lavoisier's theory of

o xi dation, and p:anck's theory of black-body radiation, to name Cases. ler it is just the eventual nossibiiitv of

se-parating a theory trom a g:ven research tradition which °IVCS .. .

the nuucal: mnressian that theories exist IQgpdent1 of

aby: c to, research traditions. This proceSs of theory separation is a fascinating one and

to be studied in some detail. I shall limit my remarks here to pointtn out that the separation of a theory Irorn its

qnT ICSCJCt I! tradition genera v takes place only when that

her th"enjr,' can I taken over, dtinfLtP Jsco 1

ficaLoua t, by an nhe,ictive research tradisioi. Theories can exist ott their own, and even when the' do it is only çuo reasons for this are clear. io: 0

Th:rtcs are neser self-authenticating; they idrialy make 7y otoioieds about the wer-ki for which they provide -P - p :asicnaie. Since 7L is one of the functions or a sgf1tradition to ;=vide just such a rationale for a theory, it is normally the ease :hatatheorvkseparated from one research traditJgpsoJy if it can so osoioed (i e jjsti ed) soccessfui rasearch tradition.

The doctrines of early thermodynamics, to which we have rcferrcci earlier, are a case in point. Originally developed within a caloricist research tradition (based on substantial, nonkinetic theories of heat, by Carnot and Clapeyron, the theory of thermodynamics proved an embarrassment during the late 1840s and 1850s, by which time the research tradition that inspired it had been largely discredited. There was wide agreement that the theory of thermodynamics was worth preserving, but not (many felt) at the price of subscribing to the research tradition which had generated it. At the same time, the kinetic, anti-caloric research tradition was making great strides

FROM THEORIES TO RESEARCH TRADITIONS 95

forward in other don-mains, but was thought to be weak insofar as it had been unable to equal, within the area of thermo-dynamics, the Successes which its competitor, the caloric

tradition, had achieved, it was Rudolf Clausius, writing in the 1850s, who was abla to show, that the theory of thermodynamics could be developed and rationalized within the kinetic tradition, independently of the caloricist assumption of the conservation of heat. Clausius thereby showed that the theory of thermo-dynamics was not inexorably wedded to the caloricist research tradition and could be absorbed by the kineticist tradition, in one fell swoop, Clausius thus managed to strengthen the case for both thermodynamics and for kineticism, by removing what had been a serious conceptual problem for them both. In like manner, Newton (as a vehement opponent of the Cartesian research tradition) was able to show that his own research tradition could absorb the I-Iuygensian theory of impact—a theory which had orig:nally been developed squarely within the Cartesian tradition.

The multitude of cases one could cite of this process ought not lead us to underestimate its difficulty. Precisely because a research tradition plays an important justificatory role for its constituent theories, any alternative research tradition which is

to play the same role must be sufficiently rich conceptually, and its partisans sufficiently imaginative, to allow it to justify and rationalize theories which prima facie are more naturally related to very different metaphysical and methodological traditions. (I shall have more to say later about this process of "theory appropriation," for it is one of the most important ways in which new research traditions establish their scientific credentials,)

The Evolution of Research Traditions

Research traditions, as we have seen, are historiga re"I

They are created and articulated withinaparticular intellectual Y a' In he generation of specific theories—and like

Just as surely as research traditions are born and thrive, s5 they die,

and cease to be seriously regarded as instruments for furthering

FROM THEORIFS TO RESE,1 RCH TRADITIONS 97

traditions have a rigid and unchanging set of doctrines which identify and define them. Any change in those doctrines, it is

I search tradition. Since, LR1OS argues, we dafine a research tradition or research Programme in terms of its central doctrines (doctrines which Lakat:s argues we make true by fiat or by convention), any change in those central tenets is de facto the abandonment of the research tradition a'hich was defined as the set of those tenets. As trmpting as this approach is (for, if true, it would make the process of ideiit7bing research traditions relatively straightforward), I shall be arguing that we must reject it, for it can on:y obfuscate our effort to get some understanding of the historical processes of seE ence.

If one looks at the grt research traditions in the history of scientific

Cartesianism, Darwinism, Newtonianisn1, Stahljan cl

lernistry, mechanistic biology, or Freudian psycholog', to flaille only a few—one can see immediately that there is scarcely any interesting set of doctrines which characterizes any one of these research tradi-tions throughout the whole Of its history. Certain Aristotelians, at times, abandoned the Aristotelian doctrine that motion jn a void is impossible Certain Cartesians, at times, repudiated the Cartesjaii identification c matter, and xtnion, Certain Nø\v tonians at times, abandoned the Newtonian demand that all matter has inertial mass. But need it follow that these seeming "renegades" were no longer working within the research tradi-tion to which they earnestl' claimed to subscribe? Does Thomas Aquinas cease to be an Aristotelian because he rejects portions of Aristotle's analysis of nlDtion? Does Huygens become a

non-Cartesianbecause he admits the possibility of void spaces? n Certain advantages will accrue if we can plausibly answer these questions negatively. To show how that is possible is the task before us.

A research tradition, w have said is a set o sssumptiiik

assumptions about the basic kinds of entities in the world,\ asStlmptons about how those entities interact, assumptions about the proper methods; to use for constructing and testing theories about those entities. In the course of their develop-nient, rssearch traditions and the theories they sponsor

/1 96 FROM THEORIES TO RESEARCH TRADITIONS

;Ik ;rocss . scine. I laE1 consider below how research raicas are displaced by other ones, for the aetiology of

. p research tradition "decav and 'putrefaction" is crucial to the

p processes which must he understood. For HOW, however, I want

It Il t a I k abou-1 bewav:a wacn important and substantive :ans can occar w[rliiiz anoioing research tradition. These hartges rake tao distinct forms.

. . I'll I The most obvaus way in I j5 by ; r'Iod!Icaizo;z 0/SOnIC Of itssiihordiizareypyjçjjjrjes

tI are Continuously I ndutyo Jucsothis iypy Rssearchers in the tradition often discover that there is,

. oim:a t the U adition, a more effective tneorv - -------- - I sonTt toe pnenomenain toe domain Alan

I a, realized pr;oi Si th alterations j?j2us tbe::es madifiatians of bourdary conditions. tCVl5iOfls of

NO xpansiaabftiic classificatory network ofa theory p1lcpir pass these are but a few of the

many ways in which the sciintist may seek to improve on the S

arabemsoh ing success of an w of the theories ithin the 7 . I research, tradition. 'Whenever he discovers a theory which is a

I::1Pcaat 011 lts predecessor he drops the latter

/ :mmediatciv. Precisely because a scientisds cognitive loyalties are casca primari: III tile researcn 'tradition rather than in any of I ts specific theories, he generally has no rational vested interest in hanging onto those individual theories. (It is for just this reason that most indi vi d ual theories have very short

. hit-spans—in many cases amounting to no more than a few racaths or even a eaks.5 Because theories change so rapidly, the

• him try ef any fburihing research tradition will exhibit a long • ;:rTsFh Tof

• tnereis another Important way in which research traditions evolve: this second class of changes involves, not the specific theories Within the research tradition, but a change of some of its most ba eiemen I must discus s this type oi transformation in some detail, since there are many philosophers who have denied that research traditions are capable of any significant internal modification. Both Kuhn and Lakatos, for instance, usually suggest that entities such as research •

/


counter a number of problems: anonialics are discovered -, ;si: oceia ErOhemS arise. In some cases, proponents of a rea: raon vIi iiflCi themselves unable. by modifying spec,fic thlcoriles \vithi:i theiradition. to eIjl1inate these anom-

luUS and conceptual problems. In such circumstances, it is conninion fur partisans of a aeseah tradition to explore what S Ls of ljilirin-ialj Iecaa made in the g

r cowl say of that research, tradition to ejtutatite the anwcahes and conceptua problems co,nfrontingdts constit-aunt theories. Sometimes, scientists will find that there s no an; cant or a :;herii;g with one or another assumption of the rocoarch tradition hich ',ili eliminate its - anomalies and a v conceptual Problems. This becomes strong grounds for aban-cn:nu sue research tradiston 'prov:ded there is some alternative

s:ght Bet, perhaps more often, scientists find that by imr:ducln - ne or too modhications in the core assumptions of the research tradition, the'; can both solve the outstanding aaameoes anti conceptual problems and preserve the bulk of thu assumptions of the research tradition in tact.

I:; she ,attcr case, it is positively misleading to speak of the creation of a Thew" research tradition, for such language oc:;cels

lnctn, us the crucial caneepivai ancestry and similarity

L s a should speak, Latncr, of a natzural a; 'dut, in hi t ie research trod/don; an evolution which repre-sents a chaae, - to be are, but dchanethat is far from rudiass a f a farmer research tradition and the creation of a new one.- '

Thore ls much continuity in ag,,ga' ''ts,tuarh tradition; - From one smae no the next, there is preservation of most of

IsS L ono research traditions pr:bun;-to/vtat technic sara ç' onwiJLdae preserved throuh the ;volUflOfl. Thoieipnrtaneeof4he'empirical prabeb;so hieb the research tradition addresses will remain

the same B t the LIpslas s hene ,

Pus1 he on O'2e coatinuic' bet-ween successive stages in thee\ogpy'

s.Li rescanch, tradition has undergone numerous evoboions in the course of time, there will probb,ly. be man discrepancies between the methodology and ontology of .its

FROM THEORIES TO RESEARCH TRADITIONS 99 earliest and its Iatet[foenua,aon Thus, the Cartesianjsni of a

ernouili, writing a century after Descartes' death, is very different from the Cartesianjsns of the master. The Newtonian research tradition in Michael Faraday's hands is a far cry from that of Newton's first followers. But a finer-grained analysis of the historical evolution of these research traditions will show that there was a continuous jflteljtuapgo Descartes tcb,BivnmeUi,. and from Newton to Faraday, and 'Tht4 the Cartesian and Newtonian research traditions, as different as their endopinnt,ay look from their beginnings, exhibited enormous IcEliltiLl

But such''d" approach leaves itself open to the obvious challenge: if a research tradition can undergo certain deep-level transformations and still remain in some sense the "same" tradition, within a research tradition from rçpJaement of one research tradition by another?

rtial answer to the question comes from recognizing that

are more central to, more entrenched within ti idition than other elements lih are

ke at that tuiPie to be rnpsL çh,a. Lth research trjj9n, To aimotion them i5 indeed to move outside the research tradition, whereas the less central tenets can be modified without repudiation of the research tradition. Like Lakatos, then, I want to suggest that certain elements of a research tradition are sacrosanct,

without repudiation of the tradition itself But ujjjnL .i t ItZbiIsou'sjflhiat the set of c/cozen sjblligg in U; is reject able) class changes throu,gh time What was taken to charac terize The 'tiijectaiaje core of the Newtonian tradition in eighteenth-century mechanics (c. ., absolute space and tinic) was no longer regarded as such by mid-nineteenthcentury Newtonians What constituted the essense of the Marxist research tradition in the late nineteenth century is substantially different from the "essence" of Marxism a half century later. Lakatos and Kuhn were right in thinking that a research programme or paradigm always has certain nonrejectable

JO;

FROM THEORIES TO RESEARCH TRADITIONS 101 Research P radii ions and Changes in Woridyjews

We have stressc'd, both here and in the Previous chapter, how research traditions and theories can encounter Serious cognitive difficulties if they are Incompatible with certain broader systems

constitute conceptual problems which may Seriously challenge the

we appen at i7Ji75'sziccessjjTresearci tradition will lead to tile bandon- IliClit 0,, that worloiriew it'hicii is ill.COflipQtjb/e

with it, and to

the eThboration c; a new worldi.'iew conipait Ic with the re-

that many radiZaT5' new scientific systems eventually come to be "canonized" as pct of our collective "common sense." In the seventeenth and eighteenth centuries, for instance, the new research traditiorw of Descartes and Newton went violently counter to many of the most cherished beliefs of the age on such questions as "man's place in Nature," the history and extent of the cosmos, and, more generally, the nature of physical processes. Everyone at the time acknowledged the existence of these conceptual problems. They Were eventually resolved, not by modifypgj offending farch traditions to bring them in line with more tradiriona vievs rathr by forging a

scientific research traditions" InTa'ocessf 're-adjustraent occurred and Marxist research traditions in

the late nineteenth century; in each case, the core, "nonscien-tific" beliefs of reflective people were eventually modified to bring them in line with highly successful scientific systems.

But it would be a mistake to assume that svorldyiews always

research traditions which

em as to dismiss fluff, The history of Science, both recent and

distant, is replete with cases where woridviews have not evap-orated in the face of scientific theories which challenged them. In our own t:me, for instance, neither quantum niechan-ics nor hehavioi'istir psychology have shsfted most people's beliefs about the world or themselves. Contrary to quantum


ents associated with it; bat they were mistaken in failing h

to at the claments constitutin fhi' class can shift through By relativizing the "essence" of a researcb, tradition with

to ume, we can, belteve, come much closer to

jr in tho way trt wh:ch scientists and hIStorians 01 science ; utilize the concept of a tradition.

I cou:se, this stib ,eaves ananswerea how it is that scientists Ic at aIO' btiel; time ss itich elements of a maxi-theory or

trah:t:on are to be treated as "unrejectabie" (a em likewise unanswered by Kahn and Lakatos). I cannot a fily satisfactory answer to the question, but some hcs are probably 'sorth exploring. Both Kuhn and Lakatos to believe that the decision about which elements of a

i-theory fail into this privileged class is arbitrary and not coca ov rattond considerations: on their account, it simply 'pens. am unable cc give a full specification of all the rs 'vhich tnfuence the selection of the core of a research

ition, but there are dearly dimensions of the choice which raucna.. For instance, on.ewsf catrenchmens of any cement of a research tradition is its

0 Lyuww.c'thzess. s of any is research tradition are continuously undergoing conceptual t(r. Same of those asttfIBti at given hme, be

id so he strong. and unproblematic. Others will he regarded 'ss ciear, ess web-founded. ments en tess. or cast doubt on, different elements of the research

entrenchment of the different ponents ,, ill shift. During the evolution of any active rrch tradition, scientists learn more about the conceptual osdence and autonomy of its various elements; when it can aowtt that certain elements, previously regarded as essential e whole enterprise, can be jettisoned without compromising problem-solving success of the tradition itself, these dc-

ts cease to be a part of the "unrejectable core" of the itch tradition. (For instance, after Mach and Frege argued none of the other elements of the Newtonian tradition

ired the absoluteness of space and time, these notions ed perceptibly towards the periphery of the Newtonian arch tradition.)

FROM THEORIES TO RESEARCH TI1ADITTONS 103 This position is had philosophy and false history. It is philo-sophically weak insofar as it ignores the fact that there is no reason, in Principle, why an entrenched worldview could not Provide a more convincing rationale for an innovative theoret-ical development than for a traditional theory. Boring's claim that a Zeitgeist automatically favors traditional theories is thus without cognithe foundation. The view is equally misleading historically. It is well known, for instance, that the Zeitgeist of late seven teentheentury England did much to hasten the replacement of she older mechanical philosophy by the newer science of Newton, precisely because Newton's research tradition could be more readily justified within that framework than the mechanistic science of Descartes could. More recently, the emergence of "new" quantum mechanics in the late 1920s found a quick and ready reception among the many intellec-tuals who were already convinced that the rigid causal cate-gories of classical science were unreliable.

The Integration of Research Traditions

Up to now I have spoken as if research traditions were invariably in competition with one another, suggesting moreover that the resolution of such a conflict coni wheo One alone among the competing traditions dominates and when its competitors are effectively vanquished. This is often the case. But it would be a serious error to assume that a scientist cannot consistent y wor in more t an one rdsear tra i ionl f these11

refcrffE1lonsare1nconsitiundamentals,then the scientist who accepts them both raises serious doubts about his capacity for cjear thinking. But there are times when two or more research traditions, far from mutually undermining one

briefly here. J'c"i"L:S 01 such situati ns which i want to discuss

There are basically two ways in which different research ,e niteggrated. In -some cases, one research

tradition ca-- another, without any serious


chanics, most people still conceive of the world as being pulated by substantial objects, with fixed and precise prop-ics; contrary to behaviorism, most people still find it helpful

tall: about the inner, mental states of themselves and others. Confi'nted hit such cxan;pics. one might claim that these tearcit traditions are still now and that older world views cdom;nate mlv because the newer insights have not yet pene- tIed ftc geacri consciousness. Such a claim may prove to he tract, bui before we accept it uncritically, there are certain Ore str:k:n historical cases that need to be aired. Ever since

o seventeenth century, the dominant research traditions thin the ph-;sica sciences have presupposed that all physical mnges are stftject to invariable natural laws (either statistical

ncosstatist:ca. Given certain initial conditions, certain

insequences oud inevitably ensue. Strictly speaking, this

dirt should be as true of man and other animals as it is of ars, planets, and molecules. Yet in our own time, as much as

the seventeenth century, very few people are prepared to andon the conviction that human beings (and some of the

igher animals have a degree of undeterminiation in their ttfuns and their thoughts. Virtually all of our social institu-ons, most of our social and political theory, and the bulk of

it nnonal Philosophy h si1i hscd on a wor1viOw seemingly

toompatible n -th a law-governed universe. Despite repeated [forts in the last three centuries to explain away this conceptual tublem, it is fair to say that this is one important area where to tradition: woridview has made very few concessions to the broader implications" of some highly successful scientific

aditions. It has long been fashionable to imagine that the woridview or

/eitgeist" J any epoch always plays a purely conservative role,

oppressing intellectual innovation and encouraging the reten-ion of the scientific status quo. Exponents of scientific progress iase frequently bemoaned the use of "worldview" considera-ions which invariably stifle the emergence of new scientific ticas. E. G. Boring spoke for many scientists and philosophers then he insisted that: "Inevitably by definition the Zeitgeist

ovors conventionality . . . [and] works against originality.""

pam: THEORIES TO RESEARCH TRADITIONS FROM THEORIES TO RESEARCH TRADITIONS 105 compromise

between the research traditions of Cartesian and Newtonian physics. In the eighteenth and nineteenth centuries, geological followers of Hutton were hammering out a new tradition which drew on elements of caloricist heat theories and Vuicanist geology These research traditions could not be Preserved intact and, as a result, the Huttoniais had to forge what WHS regarded as a 'IevolutionItry" research tradition which

incorporated elements of traditions which had been previuusly incompatible Within economics, Karl Marx drew on elements from the idealism of Hegel, the materialism of Feurerbach, and the "capitalism" of Adam Smith and his English followers.

"Nonstandard" Research Traditions

It would be dishonest to leave the topic of research traditions without adding a caveat, although how important it is remains to be seen. We have thus far characterized research traditions as rather ambitious and grandiose entities, replete wit]] ontol-ogles and methodologies. There is no doubt in my mind that

research traditions in science possess both these characteristics But there also seem to be traditions and schools i ii

0-1— ~01 if' sonic cases Ot ), have nonetheless had a gellujile D science w'hicIthouglflaekj1, One or the

intellectual coherence about theni. For instance, the tradition of

century seems to have been held together by little more than the conviction that mental phenomena could be mathematical!:, represented, Equally, the tradition of rational mechanics in the eighteenth century Seems to have

cut horizontally across almost every conceivable n]eta-physical and methodological tradition and to have drawn together a group of thinkers committed simply to the mathematical analysis of motion and rest. The in tradition of "analytic physics" in early nineteenth century France (including Biot, Fourier, Ampêi-e, and Poisson) seems to have had no common

ontology, although its partisans doubtlessly shared a common methodology. In our own time, cybernetics and information theory seem to be "schools" without well-defined

lification it; the presuppositions of e:ther. Thus, in eigh-were simul-

tousiv Newtonians and subtle fluid theorists. Their adher-to the research tradition of subtle fluids (which was as

:1; Cmoesie; as it was ewtonian) led them to postulate

ereeptible aetherial fluid; in order to explain the phenom-of emr:cy. maqnetsn;. heat. perception, and a range of

tr emmrica- prbems. Their Ne;vtontamsm, on the other

ed them u assume that the constituent oarticles of such

Js interacted not by contact, as the Cartesians tried to vest) but mther by means of strong forces of attraction and lsion, acting-at-a-distance across en;ty space. The fusion

hese two research traditions was to constitute itself a major 'arch tradition, one which Schofield has labelled "material- ,arc

endermtnag the presuppositions of neither of

Imedecessot's, the amalgamation suggested important new

a of research, and put scientists in a position to deal with

p:ncal and conceptual problems uhich neither of the

estor traditions alone could resolve satisfactorily.

a mher cases, however, the amaigan'ation of two or more ,are'; traditions requires the repudiation of some of the

ZPiRdifdP blddiEnth5f edcbi of the ffR'PB'seu;gcontined. S r1eW re —requires

sea Conneat 01 its 2ieeLLssot'5. L Is, iIieiUehiui1, iii J ust 7vcititions take place; not

thearticulation ofeseareFPtiaditiois whose inrcdieiits are

eareb tradition whose novelty consists in the way in which

ingreGnents are combtned.j [here are many examples of this

cess ididhistor of any discipline, scientific or otherwise. censider some scientific cases first, eighteenth- and nine-

nth-century natural philosophy is repete with such integra-us. Roger Boscovich, for instance, set out deliberately to /elop a new "system of nature," by picking and choosing In among the assumptions of two incompatibie research ditions, New tonianism and Lcibnizianisrn. Maupertu is at-npted something similar. The work of their contemporary, niel Bernoulli, illustrates an analogous attempt to forge a


ogles. Whether, on further investigation, it will turn out

these nonstandard" research traditions do have ontolog-sad nliethodollolgical elements or whether, failing that, they

Jmcreaus front "richer" research traditions are anaas'vered questions. Much research is still needed on

units that are too narro'v to be full-blown research

tlCrtS hut t: elobal to be mere theories.

The Evaluation of Research Traditions

sr focus titus far has been on the temporal dynamics of tab traditions. We have learned something about how such dons mu a. how they interact with their constituent des and with 'aider elements of the vorldview and the 1cm situation. wever. I have said nothing yet about how, if at all, it is bile for sc,entis:s to make sensible choices between alterna-

-eseareb traditions, nor about how a single tradition can be

mmd rehlttvc to its acceotabilitv. This is a crucial issue, for and unless we can articulate workable criteria for choice

ems sire larger units I am calling research traditions, then ave neither a theory of scientific rationality, nor a theory of

restive, cognitive growth the text few pages. 1 shall be defining some criteria for the

sation of research traditions, and discussing some of the

rent contexts III which cognitive evaluations can be made.

Progress

n though research traditions in themselves entail no

able consequences, there are several different ways in they can be rationally evaluated and thus COM

nodes atf_al at 11ever, are ost decisivb. One of these modes is synebronic, flieOther is soeid develop mental. ifOITf1Thi'gR' about the (momentary) adequacy

research tradition. We are essentially asking here how V(-' , the Iarp.ct theories within the research tradition are at

ng problems This, in turn, requires us to determine the 1cm-solving

fectivelless of

FROM THEORIES TO RESEARCII TRADITIONS 107 constitute the research tradition (ignoring their predecessors)

e we CUICittly iscusse sow to evaluate the Problem-solving effectiveness of individual theories, 17 we need only combine those appraisal's to find the adequacy of

the broader research tradition. Alternatively, we may ask about the progressiveness of a

research tradition. Here our chief concern is to determine sether the resea ad t' I

increased or decreased the Problem-solving effectiveness of its

story of 'the reseat ch tradition, we car,

L 0 subordinate measures U, general rubric,

lyLUculai important:

general prooress o deter search

mined by corn arm the'ade will ich constitute the oldest and iich constitute the most

tradition' of a research traclit* changes lon—here, the

in the molilentary adequacy of the rcsearST~i —tra during any speci ie tine span are id dition

o ro ress o and the rate a researe trdjtjon ma' be widelat odds For

general progress ins ance, a researe tradition may show a high degree of , and yet show a low rate of progress, especially in

its recent past. Alternatively, a research tradition may have a high

rate of progress during its recent past while exhibiting limited general progress.

Likewise, and even m pp ore importantly, the araisals of a

research tradition based upon its pr oressiyentherejieral or time-

depeildent) may be very different from those based on for example, where the adequacy of a research traditi high an o is relatively ti yet it shows

ie no general progress or even a negative

rate of progress. (In fact, many actual research traditions have this character.) Alternatively, there are cases (e.g., behavioristic Psychology and car y quantum theory) where the general

r.IoNs FROM THEORIES TO RESEARCH TRADITIONS

:

FROM THEORIES Q RESEARCH TRADITIONS . her the adequacyof_the tradition is still

- 109

these cases, the Sci~ntist must hue V. tc;eess

. .

Commit h—, tively to the _ ..

t iwa's toint in . appraisals ;\111 riOt

acccp-,ane, traditions and to the they can (and (1it5

directions, b tile very ac rcjectlojj oi ot1ers.

s h er rs C e to ow, call I

make a conerent decision? There are a wide

ane of poss answers U e O O e S 1' 0 C eOfl U ulS5S of research 1 i ductl\ists say choose the theory with tile highest degree tinadin iors are made. It s that issue Y,,hich must occupy us next. of corifirn, 1c i or choose the theory with the highest

' , I I I A em wse. The ; - etancc and Pursuit

ttlityfi1 i 0 stsif they give any advice at all—will say 'choose the theory with the -

S i standard i fl S On scien tifc appralsai. degree greatest of fa1sifiahi1j" Still others, such as Kuhn , would insist

. .

that no rational choice discussions s ef H:her AC it Ck

to Pn:!oS:es. 39 . can be made. I have already Indicated h none of these 1rsers

SL ace C t 3 rG' 1 features:1v assume that there is are satisfactot y own reply to the question, t) C C Wh i ch iltCOLS Ctl be of course, would be,

'thetheoiy (or research Ili:,heSt

1 a e a SSJr r context has o 00 s ib ----

_) this

the amp1reu cii scientic _ he rationale oi accepting oi reJecting any theory is thus theories. Beth these assuml:tions abIv need to be aban- 1enta11y based on the idea of problem solving - solving If tL)ned; the first because s false, second because it is research tradition has solved tor imited.

irliportarit more problems than its rivals, then accepting that tradition I t' that a Ldi I

rational precisely to the degree that we are aiming to Drooress - ""

to maximize the s pe of solve Problems. j other sor

. .

TIC I and research t AP1& s, one

I contexts of a

very ddferent sorts q tions are raised about the tazcc2osLflt,lt1O i a better problem sofve/ thqjt wa/s cogui:ive credentian of a and that much scientific

tt This way of appraising

which appears rrai1. we insist on a uni-con- research traditions has three distinct adYantages —cm e perue en as highly rational it se lbs previous models of e

-

Un 1ebotinuctijfand for hedivct-2entoasoitne,. ff euatjon measures seem (at least in ilincp le

c Erinning wita the more iarnl ' di' to ose 'er

, i SnnL taneous y offers an account of ratiozi I

LU scientists often choose to accept floSfl i ic pi o1ess ss ici s ows e o o e li cu togetner III -I ig hc esatd"eseh S

-; - .. atUasiitnu f Y applicable to the actual W, certain experiments or practica l actions are Co m -

plated, this is the operative When for instance, a The prescribe medication for a volun- research immunologist must Prstsicist

Context of pursjt. Even if we had an adequate account of y choiceithin the teer in an experIment, when a1 p decides what measuring context of acceptance, however, we would still be instrument to use for studying problem, when a chemist is very far from Possessing a full account of rational

appraisal. The reason for this is d with certain properties; in all seeking to synthesize a compoun \si i flint there arc many important situations where scientists evaluate conlpetiiig theories by

:t

f ; I FROM THEORIES TO RESEARCH TRADITIONS FROM THEORIES TO ROSEARCJI TRADITIONS

criteria which have nothing directly to do with the acceptability in this essay that I 11'nurn number of

T :1 Led :::JIII : the theories question.

beensuch situations

generation o a minimum conceptual problems and n malies _s tile

ObseoTci PJ Feverabend. it: particular, has identified many Central aim of

investigate ae seefl tnat such a vIew •

accept aran n-sioncal cases 0

here scientists he dandpursued

mcories w heinselves to be tile "lost successful problem

able, less rthy of belief, than their rivals. Indeed, the But neccl the acceptance of a -given eLergt : :o r2Ia1i' cver new research trathonoccursuiider tradition

Just such c:rcuinSt0000S. \V nether we look to CopernicanisnL -re ]"consistent with it?

A Under certain philosophy, the atomic the

es 1 011 Is cidedly theor: i the first half of the nineteenth century, early psycho- see why, we

L:iytk heori. tho preliminary efforts at the quantum mechan- we have two competing research

j. of se: suppose

" po - a: :0 moiecu!ar structure, we see he same pattern: suppose furth tI i 1OflS R and RT';

begin . (_I IIIC monlentarv 'id

s:eois: utO to 1)USUC and to explore a new research . lwher than that f ' ' equac o RT much

- grter than di j , that the rate of progress of RT' is _____ - _ -- . . \ I

related ud value for RI' c ) soppUit OIftS its novel predictions COflCCFIICCI, RT is ' '

0 ;i as (JCCCJ)taflce is v- . . . I clearly £eOflI acceptable çu::es 1 to DO accepted over its older, more successful rivals. / may nonetheless d

one o the pair. We C ess eclde to work •

Another sdc to the samo coin is the historical fact that a ' explore th • ' further articulate and

- L prooicnsoj11 merits ' :cicatist can often and grounds that it has recently

RT , precisely on the

OlVenI, siste ud1r1onspaItIculrl\ gencran g ne SolUUofl:tobh itself to be capable of

pt

a r s of sceo f iut or it s commoob the is particularlyCiflS at an impressive late This

: . his . approprilite if RT i • •

case tnat a sc:enus: vIii spend part of his time working On the traditioii It a relatively new research

. . , . . common knowledge that ntna1t resarh traddion and a cart

o f his time workino on traditio b IO5L new research

0 1 g iicw tlflaiyte tntJ • ' .- •: one or more f its less successful, less fully developed rivals. If on the solution of problems.' robleni TI

techniques to bear

we take the view that it is rational to work with and explore . (in the cliché) "fresh approaches" new techniques constitute

iy the theories one accepts (and its corollary that one ought short run, are likely to which, particularly over the

pa not accept r believe mutualy accept a budding research

problem-solving divideids. To

/ can he no va o making sense of this common phenomenon. high rate of progresswould,a ition merely because it has had a

He Le OLL be1 c ise of m Itua1 h incoos stent them os noi s old be ull of course be-a mistake out it

tbevestigrresssuccessfuitheories—bothweil-attested exhibited at 0 refuse to pursue it if it has

insist that the— conceptual) which • solve some problems (empirical or

its older, and context of acceptance exhausr scientific rationality. Confronted rivals hiv f- i 1 .

generally more acceptable

4 ------- ; C ci 1 lIt to soye . V by such cases, we wou!l have to conclude, with Feverabend Putting t • •

/ and Kuhn that the history of science rational to urLz ' enerall ' we can say that • • ays

; if on the other hand, v,c realize that scientists can 7aie good1 oJ pro,ess reasons foi %o1kiUO on not if the former has a lower *-- _ --------. %ng eIIecLivefless .

this fiequent phenomenon ion may be more compichcnfiblc suing such research traditionQIPtcinioves for put

To see what could count as "good reasons" here, we must we

return to some earlier discussions. It has often j cested become nioFe successful ' ituither development, RT' could

T v e might have grave doubts

it FROMTHEORIES TO RESEARCH TRADITIONS

422Ct RY ever becoming generally successful, but feel that some of its more progressive elements could eventually be incorporated within RT. Whatever the vagaries of the individual case, if our general aim is increasing the number of problems

e cannot he accused of inconsistency or rmtionahtv

111 1111"C pursuewithout accepting) SOfliC highly

grgressive research traditi n, regardless of its momentary madequac fn the sense defined above).

arguing that the rationality of pursuit is based on relative progress rather than overall success, I am making explicit what has be cesct oc in scientific usage as pron ie or fecpdigy, There are numerous cases in the history of sciencd1

the role 'vhich an appraisal of promise or prcgressh cross can have in earning respectability for a research.) tradition.

The Galilean research tradition, for instance, could not in its carlv years begin to stack up against its primary competitor. Aris:otaianism, Aristotle's research tradition could solve a great many more important empirical problems than Galileo's. Equahv. for all the conceptual difficulties of Aristotelianism, it tead'; posed fewer crucial conceptual problems than Galileo's early brand of phvsica. Copernicanism—a fact that tends to be 1Ott fligh) of in the general euphoria about the scientific revohatirn. But what Galilean astronomy and physics did have going for it ',' as its impressive ability to explain successfully sonre web-kaco n phenomcr.a which constituted empirical ortonialics for the cosmological tradition of Aristotle and Ptolemy. Galileo could explain, for example, svhv heavier bodies fell no faster than lighter ones. He could explain the irregular-

es on the atrrface of the moon, the moons of Jupiter, the ohases or venus, and the spots on the sun. Although Ar'istote-ion scientists ultimately were able to find solutions for these phenomena (after Galileo drew their attention to them), the explanations preferred by them smacked of the artificial and the contrived. Galileo was taken so seriously by later scientists of the seventeenth century, not because his system as a whole could explain more than its medieval and renaissance predeces-sors (for it palpably could not), but rather because it showed promise by being able, in a short span of time, to offer solutions

FROSI THEORIES TO RESEARCH TRADITIONS 1 13 to problems which

constituted anomalies for the other research : traditions in the field.

Similarly, Daltoniait atomism genera t ted so much interest in he early years of the nineteenth century largely because of its

scientific pmise, rather than it ro s concrete achievements At

Dalton's time, the dominant chemical research tradition was concerned with elective affinities. theorize about the nticr

Eschewing any attempt to oconstjtueitts of matter, elective affinity

chemists sought to explain chemical change in terms of the differential tendencies of certain chemical elements to unite W1 th others. That chemical tradition had been enormously successful in correlating and Predicting how different chemical substances combine. Dalton's early atomic doctrine could claim nothing like the overall Problensso1ing success of elective affinity chemistry (this is hardly surprising, for the affinity tradition was a Century old by the time of Dalton's

New System of G/ieiifca/ Ph1LOSOJ ) still worse, Dalton's system was

confronted by numerous serious anomalies 41 What Dalton was able to do, however was to predictas no other chemical system had done before—that chemical substances would combine

in certain definite ratios and multiples thereof, matter how much of the various reagents was present,

no

surnrnal'jzed by what we now call the laws of definite and multiple proportions created an immediate stir throughout European science in the decade after Dalton's atomic program was promulgated. Although most scientists refused to accept the Daltoniami approach, many nonetheless were prepared to take it seriously, claiming that the serendipity of the Daftonjan system made it at least sufficiently promising to be

Whethe'r the approach taken here f- 41,_

worthy of further development and l'efinenient

Tpursu1w1eyfljq}y prevail is d

1btfuI, for e have only eo to roblents in this a a;

w tat would claim is that the linkage between progress and p '

inc a ove o ers us a healthy neiddle round e One si e te InsIstence of Kuhn and the

Ves never rat nal (except in times of crisis) and the

'''I,

:4 FROM THEORIES TO RESEARCH TRADITIONS

of n_v researcl; adition—nc maflet' how rcgaesve it is—can

Aoness and the Evolution of Research Traditions

No dacrs:on of the rious appraisal vectors utilized in be complete without including the notion of

on issue often discussed under the rubric "mdc- trstah:1t " At least since the seventeenth century,

be. o:tuar. tn our own era, ad hoc stratagems and h: cthoses ha c receives much attention from scientists and

The determination that a theory or to;'eticl mcihcation is ad hoc gives us grounds, on the usual

or awrnissinC it as liegi:imate and unscientific, If we are to accept the claims sometimes made by such philosophers as Popper, Grdnhaum, and Lakatos,4 it is irrational or unscionr:fic ever- to accept a theory which is ad hoc. What does such adhocrless amount to, and why, if at all, is it such a :rabrla',- tor 7_11e4 ries which exhibit it?

Inc issue of adhocness arises most often in connection with tat e':clut;oir ot theories and the manner di which they handle anm,ies. We are usually asked to imagine a situation in which sme Aeorv, I',, encounters a refuting instance, A. in response to A, same modification is introduced into T,, producing

2": conventional wisdom insists that the later theory, T 2 is ad hoc if: 2'- can solve A, and the other known problems T cctdd solve, but 2', has no non-trivial, testable implications other than those of 2' and A. Putting it in the language of this monograph. a theory 2'- is ad hoc if it can solve only those enrpirica problems solved by its predecessor T, and those which constitute refuting instances for T, but no further problems.

There are several difficulties with this approach to adhocness. In the first place, we generally have no way of knowing at any given time whether a new theory T2 will at some later point be able to solve new problems. To make such a judgment sensibly would require a super-human clairvoyance about what empir-ical problems and what auxiliary theories (which, when con-joined with the theory, might lead to the solution of new

FROM THEORIES TO RESEARCH TRADITIONS 115 problems) are going to emerge in the future. However, taking a cue froiii Adolf Grünbauni, we can relativjze the above definition to situations of belief and say that a theory ad hoc if it is believed to solve only those eirical problems which

difficulties still remain. As Duhem taught us, iiidividual theories jjj isolation

soiution Hence, we must modify the traditional charactei'jza

tion once again, yielding a definition such as the following: a thisad hoc iJt is believed iorc esseLtiayth t solution of a/I and only those em irical rob/ejns which were solved by, or refuting instances for, an earlier theorv.

Clumsy as it is, this characterizatioti of adhocness seems to do justice to some of the most Sophisticated accounts of adhocness developed in the last decade. Assuming that adhocness is understood in this way, we are entitled to ask:

what is Objectionable about it? If some theory T2 has solved more empirical problems than its Predecessor—even just one more—then T2 is clearly preferable to T1 , and, ceteris paribus, represents cognitive progress with respect to T. However, we can go further than this to claim that the resort to ad hoc stratagems, as defined immediately above, is Perfectly consj-tent with the general aim of increasing our problem-solving

This result should not be surprising, Indeed, much of what we mean by such clichés as "learning from experience" and "the self-correction of science" is represented by situatjoxis in which, when a theory encounters an anomaly, we alter the theory so as to transform the anomaly into a solved problem. While it would be a nice bonus if every theory modification could immediately solve some new problems as well as some old, unsolved ones, to insist on that requirement (as, for instance, Popper, Lakatos, and Zahar have) is to repudiate the doctrine that theories which solve more problems about the world are Preferable to those which solve fewer.

In urging that adhocness (so defined) is a coollilive vi 'tue

FROM THEORIES TO RESEARCH TRADITIONS 117 might well wish to critieize48 do not qualify as ad hoc. The (Oh; detractors of adhocness have yet to show that the e1nendatjoi 'c- of a ffieory to preserve its problem-so1vii1 ity I nd to save it from an anomaly requires any less theoretic

capac

jl imagination or serendipity than the construction of a new theory from scratch. To the extent that these same detractors set an epistemjc premium on theories which work the first time around without any juggling or ad hoc adjustments, we are entitled to ask for the rationale for such a preference

To this philosophical worry, we should briefly add a historical

one. Most of the major theories in science—' Newton ian mechanics, Daj'inaj1 evo U ion, lvlaxwellian eectromag

netic theory and Daitonjan atonjsfl_\ere all ad hoc in the sense defined above. Those modern philosophers and scientists

adhoCness a debilitating handicap for any theory which exhibits it must explain why the most "successful" theories of the past were also highly ad hoc.

There is a train of truth, however, in the worries of many sci1tjsts and )hilosophei.s about adhocness To locate it,

ect our attention and towards the conceptual

one. in many of the c1asic epi- H"een made (e.g., Ptol-

the Lorentz- FitzGerttld contraction), the cogflit features of the situation can be characterized as follows: a theory,

T1 , has en- anornay, as een rep aced by T2 , which solves A, and T1's

previous solved problems, but is not known to be able to solve any other empirical problems. At the same time, T2

has generated more acute conceptual problems than T1 exhibited (perhaps by making assunlptioiis contrary to the ontology of T1 's research tradition or by running counter

to other acceptable theories). In such cases, the empirical gains made by T2 may be more than offset by it

uslllg to accept T2 in preference to T1 . Viewed in this light, the only legftinzat/ pejorative sense of a llreduces to a situation in which a theoi ' overall rob/'m-sol1g,7 Vlue Ot its Increasing concept'

oncep ual difficultics. Is —SO o onin

III", FROM THEORIES TO RESEARCH TRADITIONS

are (H afl(3biV bCEC ODES. Fd:1Cr.

:1S that an ad hoc 113(30Ev is preferable to its 11013-ad hoc

1CdLCSSO( confronted with known aflOI335lCS)O

beoeve othcrvise is to deny a vital aspect of the problem-r a s nUtcg

But it might be argued that I have missed the point of the cidcs of adhoness. They might say, "Yes, of course, T 2 is

Ull It'sroTute preoccessor T 1 but the relevant corn-!Da,nson is bet; sea the ad hoe T and some other theory T which is not ad hsc bur tiil solves as many problems as T2." Einstein's

theor- of relativity might exemplify T while the Lorentu-mod:tied aethee theory was T:. The obvious reply

such cr;tie:m is to ask why the admittedly ad hoc character the Lren:z contraction constitutes a decisive handicap

it in comparing it with special relativity. If the empirical iem-Ptin capacities oil the two theories are, so far as we

can te , eouvalent, then they are (empirically) on a par; defei:ders of the view that the adhocness of T. makes it di:incdv iafevior, to T, must spell out why, in such cases, the comparable problem-solving abilities and equivalent degrees of ampir:ca: support can be thrown to the winds simply by stipu-latia that ad hoc theories are intrinsically otiose.

seen-is to lie behind many discussions of adhocness is a con v :ctcn—ar:en present but t'areis defended—that there is so s, a out'-an y change in a theory which is

remove an dnomdlThe presudip-tion is tha we cannot i'kkIl EBstfBct cosmetic surgerY because, once w we know what the anomaly is, it is little more than child's play to produce some face-saving change in the theory which turns the anomaly into a positive instance. I doubt that where "real" science is concerned, this task is such an easy one. We must remember that, as adhocness has been defined, any ad hoc change must increase rather than decrease the problem-solving capacity of the theory in question. Most of the obvious and trivial ways of eliminating anomalies—e.g., arbi-trarily restricting the boundary conditions, eliminating those postulates of the theory which entailed the anomaly (assuming they could be localized!), redefining terms or correspondence rules—would generally result in decreasing the problem-solving effectiveness of a theory. Hence, such maneuvres—which we


ground for rejecting theories. But it is important to stress that the concept of adhocness itself, thus understood, adds nothing woatover to our analytic mach' for apramn theories,

since tis itself just a pqcial case of conceptual jgqlem generat:on.

by no means, the first to suggest a conceptual enterpretat:cn of adhocness: Lakatos, Zahar, and Schaffner have deve4oped similar interpretations recently.49 In all their discussions, htwever, conceptual adhocness remains but one of many sorcies of adhocness, rather than the only legitimate sense, Still tvrrse, none of these writers has indicated how conceptual adhocness is to be assessed, nor even what it entrants ic, Epuahv, all these writers leave us in the dark about

seriously, if at all, it should count against a theory if it is ad her. The seeming vi:eue of the approach taken here is that it separates spurious senses of the ad hoc from legitimate ones, and it gives us machinery for assessing the degrees of cognitive timet posed b" adhocness to the theories which exhibit it.

Anomalies Revisited

Utapter one contained the paradoxical claim that the refut-ing instances of a theory are not necessarily anomalous preblenes, a01:g with a promissory note to clarify that claim care the nuehinery was available to do so. The evaluational ntethcds outlined here allow us to return to this issue. I said before that a problem was only anomalous (i.e., cognitively threazening :r some theory, T, if that problem was unsolved bs T but solved by one of its competitors. Clearly, some re-faring instances will satisfy this definition, but many will not. It is often the case that some prediction of a theory fails to square with the data, but no other available theory can solve the data either. In the latter situation, why should the data not count as a threatening anomaly for T?

In brief, the answer is this: Whenever a theory encounters a refuting instance, it is possible to mod ify the inter1j!, associated with the theory so as todisarm the ' ' ref uti ii " data, If, for instance, we have a theory, T, that "all planets move in ellipses" and then discover a satellite of the sun, S, which

FROM THEORIES TO RESEARCH TRADITIONS 119 moves in a circle, we can always modify the interpretative rules governing the term "planet" so as to exclude S. thus preserving our theory

intact and eliminating any appearance of refutation. If there is no other theory

extant which can explain the motion of 5, the exclusion of S from T's domain is perfectly reasonable and progressive—for we lose none of our previously won problem-solving successes by legislating S out of the relevant dotnaiii. By contrast if Some alternative to T can solve S, then T's legislation of S outside the domain is a regressive step, open to rational criticism Precisely because T's abandonment of S as a legitimate problem entails that we sacrifice some of our demonstrated problem-solving capacity,

What this amounts to is that the modification of a theory

arbitrarily in order to eliminate a refuting instance is open to

criticism only if such a move would lead to a diminished

Problem-solving efficiency. That can generally be shown to happen only

if the refuting instance is solved by some theory in the domain. Hence, a refuting instance only counts as a serious anomaly

when it has been solved by sonic theory or other.

Summary: A General Characterization

Drawing together the various strands of argument developed

in this chapter, we can conclude that:

function

1. The adeauac or e ectiveness of individual theories is a 2f how aiisi1iificaiiteiiil solve, and how many important anomalies and concentual roblems tneygenera

ie acceptability of such theories is related both to their effectiveness and to the acc related research tradition. 2. The acceptabiliq, of a research tradition is deterimined the PrToem-solving . fe—ctiveness by ts la 'test theol_if"~

promise, or rattoni ursu,tabil, of a esearch tradition is efermined by the proGress (or rate of progress) it 4. Acceptance, rejection, pursuit, and non-pUrSUit constitute tlic1r'.niiye s'tances scien-tists can legitimate an

anTtheir coflstjtucmjf theories).

120


Determinations of truth and falsity are irre/evwzl to the acptabTzv ortcpursuitaoility of theories and research

2XD evaluations of research traditions an be made hPiThTh TDrhhTrutirc context. What matters is not, in some P ote sense, how effective or rnsivea on or thenrv is. but. rather, how its effectivenessornro'resiieness compares with its cnjpqjJLgs.

can now move on to discuss the implications of this model of scientific progress for an understanding of some of the central historical and philosophical questions about the cogni-the growth uf science.

Chapter Four

Progress and Revolution

The revo/utionuin can only regard his revolution as a progress in so far as he g also an historian COLLINGWOOD (1956). P. 326

The analytic machinery developed in the preceding chapters

raises numerous significant questions about the historical evolution and cognitive status of the sciences. The function of

this chapter is to examine the ways in which a probIemsolviiig approach to scientific iry can 111M o centra listorical and Philosophical problems about Science, Ile natUre__Of_

ed srientIcrevoju

uss ions can all be profitably

in terms of the Frobiem-orlellted model outlined

c Rationality

One of the thorniest questions of twentieth-century philos-ophy concerns the nature of rationality. Some philosophers suggest that rationaft 7 misu inicting to maximize one's personal utilities; others suggest that rationality consists in believing in, and acting on, only those propositions which we have good grounds for believing to be true (or at least to be

121

Chapter Three From Theories to Research Traditions · PDF fileFROM THEORtES TO RESEARCH...

Documents

Transcript of Chapter Three From Theories to Research Traditions · PDF fileFROM THEORtES TO RESEARCH...