Reflection on rules in science: an invisible-hand …tleonard/papers/Rules_in_science.pdf · Re¯ ection on rules in science: an invisible-hand perspective Thomas C. Leonard ... Imre

Abstract Can successful science accommodate a realistic view of scientiregcmotivation The Received View in theory of science has a theory of scientiregcsuccess but no theory of scientiregc motivation Critical Science Studies has a theoryof scientiregc motivation but denies any prospect for (epistemologically meaningful)scientiregc success Neither can answer the question becauseboth regard the questionas immaterial Arguing from the premise that an adequate theory of science needsboth a theory of scientiregc motivation and a theory of scientiregc success I make acase for seeing science as a kind of invisible-hand process After distinguishingdi erent and often confused conceptions of invisible-hand processes I focus onscientiregc rules treatedas emergent responses to variouscoordinationfailures in theproduction and distribution of reliable knowledge Scientiregc rules and the meansfor their enforcement constitute the invisible-hand mechanism so that scientiregcrules (sometimes) induce interested scientiregc actors with worldly goals to makeepistemically good choices

Keywords theory of science scientiregc motivation economics of science invisiblehand reliable knowledge scientiregc institutions

1 INTRODUCTION

Science is a social enterprise and its practitioners are macresh and blood peoplemotivated by more than truth and the social good That so few contemporaryscholars of science would disagree is a measure of how much the image ofscience has been transformed in the 40 years since Kuhnrsquos Structure ofScientiregc Revolutions Why then as so ably documented by Wade Hands(2001) the disarray and controversy in modern theory of science The answeris that any adequate theory of science requires not only a theory of scientiregcmotivation but also a view of good scientiregc outcomes and how they come tobe produced Even with agreement on the virtues of a more realisticconception of scientiregc motivation (and on idea that social science can aidin the understanding of science) there remain deep disagreements on theconsequences for scientiregc knowledge in particular of adopting a morerealistic conception of scientiregc motivation

The traditional image of science plusmn also known as the Received View (Suppe1977) plusmn clearly believed that science could be successful Its project after all

Journal of Economic Methodology 92 141plusmn168 2002

Remacrection on rules in sciencean invisible-hand perspective

Thomas C Leonard

Journalof Economic Methodology ISSN 1350-178XprintISSN 1469-9427online 2002Taylor amp Francis Ltd

httpwwwtandfcoukjournals

DOI 10108013501780210137092

was to reduce that success to a set of logical even algorithmic rules forgrinding out truth claims sometimes known as Scientiregc Method But theReceived View failed to o er any theory of scientiregc motivation and thuswhatever the merits of its normative case for Method had no way of knowingwhether real scientists had the right incentives to follow the Scientiregc MethodWhat I call Critical Science Studies (CSS) su ers from the obverse problem1

It o ers a theory of scientiregc motivation portraying scientists as (at leastpartly) self-interested and as having worldly goals similar to economic agentsAt the same time however CSS abandons as impossible the entire normativeproject of the theory of science CSS scholars defend and often assume`cognitive egalitarianismrsquo `the thesis that all beliefs are epistemically orevidentially equal in terms of their supportrsquo (Laudan 1984 29plusmn31) TheReceived View has a theory of scientiregc success but no theory of scientiregcmotivation CSS has a theory of scientiregc motivation but denies any prospectfor (epistemologically meaningful) scientiregc success Thus neither can answerthe important question plusmn can successful science accommodate a realistic viewof scientiregc motivation plusmn because both regard the question as immaterial

This paperrsquos premise is that the question deserves an answer that is that anadequate theory of science must o er both a theory of scientiregc motivationand a theory of scientiregc success I will not argue for this premise except asfollows We need a theory of scientiregc success because the natural scienceswhile fallible and imperfect and not immune to politics bias fashion and fadhave nonetheless been the most successful of human cognitive endeavours(Haack 1998 130) Unless sciencersquos success is denied or can be regarded as anaccident of happenstance or as inherently inexplicable itrsquos worth investigat-ing how it is accomplished2 The idea is that science studies must not onlydescribe the practices of scientists but analyse how scientiregc practices succeedor fail in accomplishing cognitive goals

We also need a theory of scientiregc motivation even if it is the theory thatscientists plusmn professional skeptics plusmn will blindly heed the rules laid down forthem by lawgiving philosophers of science If one regards scientists assometimes having interests opposed to what the rules dictate then it matterswhether the rules are incentive compatible The history of science suggeststhat the rules are frequently not observed There may be something wrongwith the rules (Feyerabendrsquos 1975 conclusion for example) or alternativelythe rules may be regne but there are incentive problems or both Without atheory of scientiregc motivation there is no basis for distinguishing amongthese rival explanations and in addition there is no prospect for identifyingscientiregcally good rules that interested scientists will opt to follow

The present task is to ask under what circumstances can a real communityof fallible èpistemically sulliedrsquo (Kitcher 1993 384) inquirers achievethe good outcomes traditionally thought to require ideal inquirers (Haack1998 98) Invisible-hand explanations suggest themselves naturally to aneconomist plusmn they are deeply rooted in the disciplinary ethos This paper o ers

142 Articles

a case for seeing science as a kind of invisible-hand process But if it is to haveany explanatory force the term ìnvisible handrsquo must be more than a label fora black-box process I emphasize scientiregc rules which I treat as emergentresponses to various market failures in the production and distribution ofscientiregc ideas Scientiregc rules and the means for their enforcement consti-tute the invisible-hand mechanism by which I mean scientiregc rules induce(partly) interested scientiregc actors with worldly goals to make choices that(sometimes) lead to epistemically good scientiregc outcomes3 Scientiregc rulesare discussed in Section 5 as three prefatory tasks precede it

First any attempt to reconcile successful science with a realistic view ofscientiregc motivation must consider the current impasse in science studieswhich means understanding why both Received-View and CSS theories ofscience regard the project of reconciliation as moot This task takes up Section2 Second any reconciliation will likely require a more pragmatic conceptionof scientiregc knowledge than the traditional conception of justireged true beliefI suggest in Section 3 the alternative of reliable knowledge a conceptdeveloped by Alvin Goldman (1986) and others Third any attempt to usein science studies the name ìnvisible handrsquo which is often a term of derisionoutside of economics must distinguish Adam Smithrsquos sense of the term plusmnunintended if planned-looking beneregcial consequences plusmn from cognate mean-ings such as laissez-faire and Pareto optimality I distinguish the di erentinterpretations of ìnvisible handrsquo in Section 4

2 THE RECEIVED VIEW THE TRADITIONAL IMAGE OFSCIENCE

Once upon a time theorists of science who were then overwhelminglyphilosophers with epistemological concerns o ered an image of how scienceshould proceed (The rest of this paragraph is taken only lightly paraphrasedfrom Kuhn 1992 4plusmn5) Nature was its starting point as was the claim thatphenomena arrive to inquirers in the forms of observations that are factsaccessible to and indubitable for all normally equipped human observersFacts were the objective basis of the laws theories and explanations for whichthey provided the foundation And while laws theories and explanationswhich interpret given facts may be arrived at variously the process by whichscientists choose among rival interpretations plusmn theory choice plusmn was said to begoverned by the data That is the facts can be said to adjudicate among rivalinterpretations of them The process of adjudication constitutes the scientiregcmethod a logical and in some versions algorithmic means by which scientistsarrive at true generalizations about and explanations for natural phenomenaplusmn and if not true at least approximations to true and if not certainapproximations then at least highly probable ones In sum Kuhn says thecentral pillars of the Received View image of science pillars on whichthe epistemic authority of science was said to rest are two `regrst that facts

Remacrections on rules in science 143

are prior to and independent of the beliefs for which they are said to supply theevidence and second that what emerges from the practice of science aretruths probable truths or approximations to truth about a mind- andculture-independent external worldrsquo (Kuhn 1992 18)

As noted traditional theories of science simply did not attend to thequestion of scientiregc motivation Received View theory is normative andprescriptive remacrecting its genesis in philosophy Here are the rules plusmn a logic ofdiscovery plusmn that correctly applied will produce scientiregc knowledge Shouldscientists opt not to observe them so much the worse for science Put thiselliptically the foregoing account risks caricature but it su ces for ourpurposes

21 The Revisionist critique of the Received View scientiregc method andmotive

History has been unkind to the Received View Beginning roughly mid-20thcentury a distinguished group of scholars plusmn Karl Popper Stephen ToulminNorwood Hanson Willard Quine Imre Lakatos Paul Feyerabend andThomas Kuhn among them plusmn assembled a compelling critique I select andemphasize two key arguments from what is a wide-ranging critique (1) thatthe Received View image of scientiregc method is macrawed plusmn ie that it is incorrectas an account of what scientists should do and (2) that the Received Viewimage of science is itself unscientiregc in the sense that it did not systematicallyattend to (nor believe that it had to attend to) the question of whether theempirical practice of science actually comported with its theory of science

Revisionist skepticism on scientiregc method begins with the nature andfunction of facts in the Received View Regarding the nature of facts Quine(1953) argued that since experience is not prior to belief facts are not prior totheoretical interpretation Facts are rather unavoidably contaminated by thetheories devised to explain them Scientists in practice have no choice but tolook for data under the street lamps of theory The problem is particularlyacute with experimental apparati the design and operation of which maythemselves depend upon theory sometimes upon the very theory being testedIf facts are theory-laden then contra the traditional image they cannot serveas a neutral court in which rival theories are tried (Kuhn 1992 4plusmn5)

The QuineplusmnDuhem hypothesis attacks the Received View image of facts onanother nearby front Even if scientists reach agreement on what the factsare there is the problem of determining exactly what the facts say regardingthe theory they are meant to confront When evidence contradicts theorywhat has gone wrong Is it the theory or the evidence And if it is the theorywhich component plusmn the main hypothesis (eg consumers maximize utility)an auxiliary hypotheses (eg rates of time preference are invariant) or aceteris paribus condition (eg nominal income is constant along aMarshallian demand curve) or other (eg only equilibrium states are of

144 Articles

theoretical interest) In Quinersquos elegant phrasing òur statements about theexternal world face the tribunal of sense experience not individually but onlyas a corporate bodyrsquo (1953 41) Thus implies the DuhemplusmnQuine argumentwhen tests disconregrm `the choice of where exactly to point the accusing regngerof refutation is ours not naturersquos rsquo (Hollis 1994 80)

Theory underdetermination extends the DuhemplusmnQuine indeterminancy plusmnthe worry about which elements of a given theory are disconregrmed plusmn to theproblem of choice among rival theories In particular when there areincompatible theories that are empirically equivalent a given body of evid-ence will be equally conregrming or disconregrming of all theoretical rivals4 Ifper these revisionist objections to the Received View facts can notunambiguously choose among theoretical rivals there is broad scope forscientiregc disagreement But since there is nonetheless scientiregc consensus evenwhen the facts do not speak unambiguously it must be the case that theorychoice has a non-evidentiary aspect Once facts can no longer carry the entireburden of theory choice the door is open to explaining scientiregc belief byreference to non-evidentiary criteria plusmn beauty parsimony political or ideo-logical biases class membership research agenda the potential for generatingfame or fortune and so on

The second element of the revisionist critique I emphasize is the idea thattheories of science should themselves be scientiregc plusmn in particular that theempirical strategies of science should be used in theory of science as well5 Thisview in theory of science is typically referred to as `naturalisticrsquo and itsproponents as naturalizers6 (Callebaut 1993)

Revisionist scholars including those most inmacruential in economic metho-dology were hardly blind to incentive problems Popper and Lakatos forexample generally regarded scientists as self-interested Popper worried thatopportunistic scientists would resort to ad hoc ìmmunizing stratagemsrsquo whenthe data proved inconvenient for their pet theories and he proposed a welterof rules aimed to forestall various kinds of hypothesis-rescuing dodges(Popper 1972 15plusmn16) So perhaps Lakatos is unfair when he criticizesPopperrsquos rule of falsiregcation on incentive grounds

Popperrsquos [demarcation] criterion ignores the remarkable tenacity ofscientiregc theories Scientists have thick skins They do not abandon atheory merely because facts contradict it They normally invent somerescue hypothesis to explain what they then call a mere anomaly or if theycannot explain the anomaly they ignore it

(1978 4)

Regardless there is the question of how the meta-scientiregc theorist shouldregard theoretical evidence that scientists have perverse incentives or empiri-cal evidence from history of science that scientists do not for exampleruthlessly falsify their own theories Economic methodologists have longnoted the routine failure of economists to apply Popperrsquos strict method of


refutation (Archibald 1967) A naturalizing theorist of science might arguethat the failure of scientists to ruthlessly falsify should be seen as evidence thatPopperrsquos falsiregcation criterion is itself refuted (or at least unscientiregc) by itsown lights But Popper is no naturalizer He does not regard his meta-scientiregcclaims as rules justireged by an examination of the history of science and hedoes not regard his meta-scientiregc claims as testable by evidence fromscientiregc practice7 (1959 52)

The idea that theory of science cannot and should not itself be scientiregcchanges with Kuhn himself trained in physics Kuhn believed what scientistshave actually done is relevant to any theory of what they should do Thetraditional image of science prescribed rigorous testing of theory by facts butin legislating from the armchair ignoring the history of science wasunscientiregc when self-applied Kuhn in contrast conceived of theory ofscience in scientiregc terms Kuhnrsquos naturalism was not new but Kuhn getsthe credit for reviving naturalism in post-war philosophy of science8 `Kuhnliberated usrsquo says William Wimsatt `not only to do history of science but alsoscience History of science is after all just science looked at after it hashappened rsquo (Callebaut 1993 24)

Kuhnrsquos historical turn was thus theoretically motivated Kuhn regardedscientiregc practice as rather di erent than what prevailing philosophy ofscience then prescribed Faced with this anomaly plusmn scientists donrsquot do whatphilosophers of science say they should do plusmn Kuhn turned to history not todebunk the scientiregc enterprise but to improve the theory of science `Whatwe mostly thought we were doingrsquo summarizes Kuhn in a retrospective `wasbuilding a philosophy of science on observations of scientiregc life thehistorical record providing our data rsquo (1992 4)

Before the naturalistic turn in theory of science the few social scientistswho studied scientiregc rules still deferred to the Received-View bifurcationbetween the practice of science and the product of science scientiregc know-ledge Pioneers like Robert Merton (1973) mostly hewed to the traditionalphilosophical distinction between study of science (loosely what philoso-phers called the context of discovery) and the study of knowledge (looselywhat philosophers called the context of justiregcation) Merton argued thatsocial factors could inmacruence for example the selection of research problems(1973 [1957] 554) But Merton was traditional in not admitting social factorsinto the determination of scientiregc knowledge arguing

The criteria of validity of claims to scientiregc knowledge are not matters ofnational taste and culture Sooner or later competing claims to validity aresettled by the universalistic facts of nature which are consonant with oneand not with another theory9

(ibid)

This mid-century deference to the authority of philosophy in matters ofscientiregc knowledge is almost poignant in retrospect and it did not last long

146 Articles

22 Critical Science Studiesrsquo view of science

Critical Science Studies (CSS) takes a variant of Kuhnian naturalism andweds it to a radical skepticism that derives from an especially strong readingof the revisionist arguments on theory ladenness and theory underdetermina-tion Science says CSS essentially inverting the traditional formulationknows nothing

Though most revisionists argued that the facts alone could not determinescientiregc belief they were also clear that as Kuhn put it òbservation andexperience can and must drastically restrict the range of admissible scientiregcbelief else there would be no science rsquo (1996 [1962] 4) If the revisionistsdiminished the role of facts in the determination of scientiregc belief CSSscholars running through a breach opened by Quine deny facts any rolewhatsoever in theory choices made by scientists10 CSS claims in e ect thatunderdetermination amounts to undetermination of theory by data

For CSS scholars empirical evidence never inmacruences theory choiceScientists may (falsely) invoke the inmacruence of data for rhetorical purposesbut says Harry Collins `the natural world has a small or non-existent role inthe construction of scientiregc knowledgersquo (1981 3) `[W]e can never use Naturersquo argues Bruno Latour `to explain how and why a [scientiregc]controversy has been settledrsquo (1987 99) Some even elevate the view thatevidence never a ects scientiregc belief into a methodological injunctionèxplanations [in science studies] should be developed within the assumptionthat the real world does not a ect what the scientist believes about itrsquo (Collinsand Yearly 1992 372) And if evidence is immaterial then appeals to dataand reason ubiquitous in science must be regarded as mere posturing`Science legitimates itself by linking its discoveries to powerrsquo says StanleyAronowitz à connection which determines not merely inmacruences what is tocount as reliable knowledgersquo (1988 204 cited in Haack 1998 102)

The CSS sense of the term `socialrsquo is perhaps now clearer It is meant as anantonym for `naturalrsquo in explaining the causes of scientiregc belief It refers inparticular to the social subset of non-evidential determinants of scientiregcbelief plusmn social class political ideology gender and so on It is this subset ofnon-evidential determinants that CSS scholars emphasize while othersinvoke non-evidential determinants not obviously social such as theoreticalparsimony or internal consistency or congruency with other theoriesregarded as true

23 Do social inmacruences necessarily undermine knowledge

The revisionist worry shared by Kuhn and familiar to social scientists wasthat the presence of non-evidential criteria might prove inimical to scienceNon-evidential determinants of scientiregc belief clearly threaten the objectivitythat the Received View said was provided by impartial adjudication by facts


The CSS view of knowledge dispenses with that worry by the route of denyingthat science could ever produce anything worth protecting in the regrst place Ifscience is entirely social then objectivity is not just unattainable it isincoherent CSS scholars thus regard the normative project of determiningwhether scientiregc practices (social and other) promote or undermine scientiregcknowledge as obviated When knowledge is demoted to belief description isall theory of science can attempt11 CSS thus begs the vital question of whether(and to what extent) social inmacruences undermine or promote scientiregcknowledge

For a brief period some important CSS scholars adopted a debunkingstrategy with a less totalizing critique During this era CSS emphasized howsocial factors worked to undermine knowledge plusmn a critique from social sciencerather than from epistemology Instead of arguing that knowledge is inprinciple unattainable they argued that interested scientists with non-cognitive goals would not produce it (Latour and Woolgar 1986) Theargument implied that the selmacress truth seeking of the idealized inquirerwas a necessary condition for the production of scientiregc knowledge This lineof inquiry was inmacruential for a time and it invoked ideas plusmn `credit maximizingrsquoscientists for example plusmn with a familiar economic ring Some commentatorseven identireged it with the economics of scientiregc knowledge (ESK) literature(Davis 1997) But this line of CSS reasoning has been largely abandoned by itsoriginal proponents perhaps because scholarship emphasizing invisible-handoutcomes shows that interested scientists need not entail bad outcomes inscience (Hull 1988 Goldman and Shaked 1991 Kitcher 1993)

Since I am here taking the possibility of scientiregc success as a premise I willnot argue against the CSS conception of scientiregc knowledge in the largeRather I point out that the two CSS debunking strategies plusmn one emphasizinginterested scientists the other emphasizing the toothlessness of empiricalevidence plusmn are at odds so that the CSS theory of science is inconsistent Inparticular the CSS epistemic stance conmacricts with its own interest-based viewof scientiregc motivation and it also undermines naturalized theory of scienceCSS included

The CSS view of empirical evidence is hard to square with its theory ofscientiregc motivation Even if scientists care only to promote their own non-cognitive goals they need to be empirical to do so If they seek to promotetheir own interests then they must attend to the relative e cacy of pastinterest-promoting strategies an empirical task But if scientists can never beinmacruenced by empirical evidence then they cannot examine history (or anyother evidence) and are helpless to advance their own interests12 (Laudan1990 159plusmn60) This paradoxical conclusion shows how extreme is the CSSposition on evidence

The CSS variant of science-studies naturalism is also at odds with or atleast made puzzling by its view of the role of empirical evidence in scienceWhen naturalizing theorists of science argue for a scientiregc approach to

148 Articles

science studies they have in mind an empirical approach (Hausman 1992221) CSS advocates the use of sociological (and other) methods in sciencestudies which is unobjectionable by itself but is clearly at odds with the CSSclaim that scientiregc belief is never determined by empirical evidence It is at aminimum a puzzling programme that o ers lots of evidence for the proposi-tion that evidence is immaterial (Laudan cited Hull 1988 4) Why should wetreat seriously the scientiregc claims of a research programme that by its ownsweeping indictment cannot be scientiregc13 A naturalized approach to thetheory of science plusmn which proposes to import the techniques of empiricalscience plusmn cannot be made coherent with a view of knowledge that denies apriori all prospect for empirically successful science

Perhaps because it substitutes ìtrsquos all socialrsquo for the Received Viewrsquos ìtrsquos allnaturalrsquo CSS looks curiously traditional at least in its all-or-nothing aspectCSS presumes as did the Received View that the social character of sciencewould necessarily undermine knowledge And in abandoning the normativeproject of theory of science CSS essentially follows the Received-Viewpractice of indi erence to whether scientiregc practices are incentive com-patible ie whether they tend to promote or to undermine knowledgeproduction14

Just as the Received View took its `science is naturalrsquo credo as license tofocus on warrant wholly at the expense of acceptance CSS takes its credo`science is socialrsquo to focus on acceptance wholly at the expense of warrant Amore moderate perspective the one argued for here regards science as havinga social character but also sees scientiregc belief as informed by empiricalevidence It regards `science as socialrsquo not as an argument-ending claim inepistemology but as a mandate for inquiring into how the social character ofscience promotes or undermines knowledge notably how social factors lead(or fail to lead) fallible interested inquirers to accept what is empiricallywarranted (Haack 1998 110)

These very di erent interpretations of what is meant by `science is socialrsquocan be illustrated by the social practice of relying upon experts Scientistsroutinely trust (without veriregcation) the opinions of people they regard asscientiregc authorities Though trust is not without clear risks it can beperfectly rational plusmn it amounts to an epistemic division of labour (Goldman1995b 746) Because independently assaying every claim that a scientistaccepts in trying to produce knowledge is prohibitively expensive the socialpractice of relying on expert authorities is an economizing practice necessaryfor the possibility of scientiregc progress The CSS scholar who identiregesacceptance with empirical warrant regards the practice of trusting authoritiesin wholly non-cognitive terms Scientists defer to authority for reasons ofsocial hierarchy say or to logroll ie to gain reciprocal favours for macratteringpowerful people Without denying that such factors are possibly in play amoderate theory of science which sees empirical warrant and acceptance asdistinguishable also inquires into whether authorities can be seen as signaling


the reliability of the claim in question The signal may well be faulty of coursefor the putative authority may be corrupt or wrong thus the practice oftrusting authorities will be epistemologically good depending on the extent towhich authorities are indeed authoritative (Ibid) Accurate or not thoughthe expertrsquos acceptance signals empirical warrant it does not constituteempirical warrant When Charles Sanders Peirce condemned as inferiorscientiregc belief arising from authority he had in mind thoughtless (orcoerced) deference to political or religious authority the kind for examplethat made Stalin-era Lysenkoism infamous (Peirce 1877) But the prospectthat considered trust may be individually rational (and socially economizing)is precisely why one wants to inquire into among other things the e cacy ofthe social processes by which some scientists come to be regarded asauthoritative

3 WHEREIN SUCCESSFUL SCIENCE RELIABLEKNOWLEDGE

Epistemology traditionally deregnes knowledge as justireged true belief15

To know something you must have a belief the belief must be true plusmn youcannot know something that is not true plusmn and the true belief must be justiregedThe `justiregcationrsquo requirement ordinarily means that justireged true beliefs arebuilt atop secure foundations to knowledge plusmn for example atop beliefs thatare self-evidently true plusmn or are logically coherent with other beliefs alreadyregarded as true Both foundational and coherentist approaches to justiregca-tion emphasize the ability to connect vertically or horizontally a given beliefwith other beliefs regarded as unimpeachable

The trouble with the traditional conception of justiregcation noted by manyis that it sets the bar too high Says David Hull

[T]he content of science can[not] be `justiregedrsquo in the sense thatgenerations of epistemologists have attempted to justify them Thereason that epistemologists have not been able to justify knowledge-claims in their sense of `justifyrsquo is that no such justiregcation exists Theywant the impossible

(1988 12plusmn13)

CSS scholars read the failure of science to achieve the impossible not as anindictment of the traditional conception of knowledge but as entailing theimpossibility of producing knowledge in general Paradoxically the CSSreading is true only if CSS scholars accept the traditional deregnition ofknowledge Kuhn puts it as follows

[CSS theories of science] are taking the traditional view of scientiregcknowledge too much for granted They seem that is to feel that traditionalphilosophy of science was correct in its understanding of what knowledge

150 Articles

must be If science doesnrsquot produce knowledge in that sense theyconclude then it cannot be producing knowledge at all

(Kuhn 1992 9)

An alternative approach used in this paper is to adopt a more pragmaticand more modest conception of knowledge I use reliabilism an approachpioneered by Alvin Goldman (1986) and others that qualireges a belief asepistemologically justireged when it reliably indicates truth One formulationsays a belief qualireges as knowledge if someone believes the proposition it istrue and if it were not true the cognizer would not believe it For examplesomeone knows that the telephone is ringing if he believes this it is true andif the phone were not ringing the cognizer would not believe that it wasringing because he would not be having the same auditory experience (Theformulation and example are from Goldman 1995a) Reliability can refer tobeliefs and to their mode of acquisition Goldman has emphasized thereliability of the processes or practices by which beliefs are formed wherereliability consists of generating a fairly high ratio of true to false beliefs (truthratio) Goldman uses the term `veritismrsquo to express the reliabilistrsquos focus onthe ability of beliefs and the modes of their acquisition to `trackrsquo or `hook upwithrsquo or ìndicatersquo truth (the adjective is `veritisticrsquo) (Goldman 1995a 199987plusmn94) Beliefs are states of knowledge error or ignorance and have funda-mental veritistic value (or disvalue) and practices have instrumental veritisticvalue insofar as they promote or impede veritistic beliefs (Goldman 1999 87)

Economic considerations enter because states and practices with higherveritistic value will tend to be more economically valuable There are clearincentives to adopt more veritistic beliefs and practices given the costs (intime e ort and risk) to so doing16 False beliefs or ignorance can lead me tolend money at one per cent while revolving credit at 18 per cent or to preferriskier investments with lower rates of return or to regard a nominally higher(but in real terms lower) wage as increasing purchasing power or to confuselegal and economic incidence of a sales tax When a credible central bankadopts a tight-money policy short term interest rates are likely to rise andwith them the cost of regnancing goods bought on credit This macroeconomicbelief has positive if limited veritistic value for the exact timing andmagnitudes of changes in spending are hard to forecast But one can befairly sure of the direction of change inexact knowledge that is economicallyvaluable to informed decision makers

More generally seeing beliefs as at once fallible and valuable has a levelinge ect that shifts emphasis from the traditional (and CSS) epistemologistrsquos all-or-nothing appraisal to consideration of the decision makerrsquos practicaljudgment under uncertainty The epistemologist who insists that beliefs arejustireged true belief or they are nothing is in practice a radical skeptic Ineveryday life such a stance would be paralyzing a non-starter It amounts toa kind of radical risk-aversion an unwillingness to accept any cognitive risk


even when potential returns are large (Rescher 1989) A scientist who acceptsthe all-or-nothing criterion e ectively refuses to enter the cognitive enterprise(Rescher 1993 86) The risk of cognitive error along with other costs ofinquiry is the price of veritistically improved scientiregc beliefs and practices

In practice all but radical risk-averters are willing to risk error time andmoney to gain greater veritistic value But they face an information problembecause what is veritistically valuable in a belief or practice may not be known(Goldman 1999 91) Veritistic value is not unknowable but some veritisti-cally valuable beliefs and practices may well be unknown In science theproblem is especially acute because the object is in part to produce novelclaims plusmn the beneregts of research can be quite uncertain ex ante (Loasby 1989197) Even retrospectively the value of scientiregc knowledge can be di cult tomeasure (Dasgupta and David 1994 490) This is not an argument againstrational consideration of beliefs and practices it is a recognition thatoptimality may be di cult to realize or even underegned Scientists like therest of us take into account the costs and beneregts of their choices as best theycan whether or not they can equate the marginal gain in veritistic value withthe marginal cost of obtaining it

4 SCIENCE AS AN INVISIBLE-HAND PROCESS

An invisible-hand process can be characterized by the following conditions(1) individual actions lead to unintended consequences (2) the aggregatee ects of individual actions result in a spontaneous order17 that gives theappearance of design by a master planner and (3) the order that results isdeemed beneregcial in ways that the individuals did not intend but neverthelessregnd desirable (Vaughan 1987)18 Adam Smith uses the term ìnvisible handrsquoonly twice in work published during his lifetime most famously (and closestto the sense just deregned) in the Wealth of Nations

[E]very individual neither intends to promote the public interest norknows how much he is promoting it [H]e intends only his own gain andit and he is in this as in many other cases led by an invisible hand topromote an end which was no part of his intention Nor is it always theworse for society that it was no part of it By pursuing his own interest hefrequently promotes that of society more e ectually than when he reallyintends to promote it

(1937 [1776] 423)

Strictly speaking invisible-hand processes can also lead to collectively badrather than good unintended consequences Prisonersrsquo Dilemma settingssuch as certain common property resource problems (grazing land riparianwater regsheries) provide an example Common property resources arerivalrous in consumption and exclusion is costly so self-regarding choices

152 Articles

create external costs that unintendedly lead to inferior collective outcomesie to overuse or destruction of the resource19

Invisible-hand arguments are uncommon and unpopular outside of eco-nomics and science studies is no exception It is not merely a predispositionamong science-studies scholars to be skeptical of or even hostile to markets plusmnthe leading example of invisible-hand processes plusmn though it is true thatscience-studies scholars come from academic disciplines sociology especiallyfounded (in the US) partly in opposition to a nascent neo-classical economicsThe problem I suggest is more one of misconception

Non-economists regnd it implausible that absent the guiding visible hand ofauthority self interested action can lead to good social outcomes Or more tothe point they tend to conmacrate invisible-hand explanations with a kind oflaissez-faire The conmacration may arise from the termrsquos close association withAdam Smith and Smithrsquos seminal role as the proponent of free trade againstthe mercantile view the original debate over the proper scope of the statersquosrole in the economy (Demsetz 1993 159plusmn60) But just as it is false to assumethat the choice is dichotomous between central planning and anarchy it isincorrect to cast the original invisible-hand theorists plusmn David Hume AdamSmith and others we might group into the Scottish Enlightenment plusmn asproponents of vulgar laissez-faire They were in fact clear that governmenthad a crucial (if limited) role to play in underwriting invisible-hand outcomesSmith conceived of his work as a `science of the legislatorrsquo that is as advice tolawmakers on how to create an institutional structure that could best fosterinvisible-hand outcomes In modern language the invisible-hand theoristsrecognized that functioning markets rely upon laws that credibly establishand enforce the protection of persons property and contracts Moreovereven `freersquo markets undergirded by property and contact laws can fail to resultin optimally good outcomes what neoclassical economics came to call marketfailure So while invisible-hand explanations can be used to justify freemarkets and are a rhetorical favorite of free marketeers invisible-handexplanations should not be identireged with laissez-faire The case for freemarkets is one that sees state regulatory cures as worse than free marketdiseases20

Neoclassical economics has contributed to the misidentiregcation of invis-ible-hand reasoning with laissez-faire by using the term ìnvisible handrsquo as acasual label for a decentralized model of pricing that is ordinarily ifconfusingly called perfect competition There is not much competitioninvolved In fact there is no rivalry whatsoever the competitive modelrefers to market and information conditions such that (maximizing) buyersand sellers behave as price takers The only factors that a ect choice areexogenously given tastes and technologies and prices which are determinedimpersonally in thick markets21 All of these factors are beyond the control ofany of the participants and of any authority (Demsetz ibid) which may helpexplain the conmacration of invisible-hand reasoning with laissez-faire


A subtler confusion results when science-studies scholars follow the neo-classical usage of ìnvisible handrsquo to refer to the realization of Pareto-optimalequilibria Wade Hands (1995) for example takes Kitcher (1993) to task ongrounds that Kitcherrsquos approach amounts to `little more than Adam Smithrsquosìnvisible handrsquo applied to the cognitive domainrsquo (1995 612) Hands usesìnvisible handrsquo in the neoclassical sense that is to refer `the optimalallocation of our cognitive resourcesrsquo (ibid 611) Adam Smithrsquos invisiblehand did not of course refer the optimality-cum-equilibrium emphasis oftwentieth-century welfare economics

Similarly Hands (1997) criticizes Kitcher for employing an epistemicwelfare economics which among other things involves risking the standardproblems of neoclassical welfare economics plusmn aggregation of individualdemand functions into market demand interdependent utilities and so onI think Handsrsquos critique has merit any naturalized approach to theory ofscience will import the weaknesses of its home discipline along with thestrengths and Kitcherrsquos neoclassical economic approach is no exception ButHandsrsquos argument is a critique of neoclassical welfare economics or of itsapplication in science studies not of invisible-hand reasoning per se22

A related example is Goldman and Coxrsquos (1996) argument in the contextof free speech against the claim that a free markets will always produce moretrue speech than will regulated markets They point out that free markets mayfail and that even ideal markets do not intrinsically guarantee more truthThe reason is that ideal markets are those that most e ciently satisfyconsumer preferences And if consumers donrsquot much value truth plusmn preferringinfotainment to hard news say plusmn then not much truth will be producedCorrect or not Goldman and Cox interpret what they call `marketplacetheoryrsquo as (essentially) neo-classical perfect competition23 Thus thoughthey invoke Àdam Smithrsquos `ìnvisible handrsquorsquo rsquo their critique of the market-place-of-ideas stance on truth production is not a critique of invisible-handexplanation

Invisible-hand explanations do not entail optimality (in the sense ofexhausting gains from trade) nor do they require decision making inmacruencedonly by price nor do they imply that `freersquo markets in speech necessarilyproduce relatively more true speech than more regulated markets

A number of science studies scholars apply invisible-hand type reasoningto science (Hull 1988 Goldman and Shaked 1991 Kitcher 1993 Walstad2001) even ìf from these diverse e orts a distinct movement has begun totake shape its critics have been the regrst to noticersquo (Walstad 2001 2) OnlyDavid Hull (1997) employs the term ìnvisible handrsquo and all use di erentexplanatory strategies What unireges these various projects is the idea thatscience is successful not because real scientists are selmacress truth seekers butbecause science is socially organized in a epistemically beneregcial way plusmnshowing how (and under what circumstances) epistemically impure scientistscan produce epistemically good outcomes My own emphasis is on how

154 Articles

scientiregc rules ordinarily regarded as mere norms induce fallible interestedscientists to produce reliable knowledge24 In particular do rules lead (or failto lead) fallible inquirers to accept what is empirically warranted

5 SCIENTIFIC RULES AND EPISTEMICALLY VIRTUOUSINCENTIVES

Let us proceed with rules in science by posing three fundamental questionsconcerning incentives in science (1) why do scientists produce knowledge atall (2) why do scientists openly publish their results and (3) why do scientiststend to produce reliable knowledge

Pioneering researchers in the economic of science were skeptical of the thencommonplace Received-View notion that `the search for knowledge is itselfthe highest social good and that any other beneregts that society might obtainare just social gravyrsquo (Nelson 1959 299) This skepticism led them to askwhy scientists produce knowledge at all (Arrow 1962) The òldrsquo economics ofscience begins with the idea that scientiregc knowledge is non-rivalrous inconsumption And because the individual scientist has trouble appropriatingthe returns to a discovery (once made public) a market failure results plusmn theunder provision of a public good (or of a positive-externality good) owing tothe divergence between individual and social returns (Dasgupta and David1994 490)

The remedy advocated was state intervention the state produces theresearch the state subsidizes research (claiming ownership of the output)or the state creates and protects intellectual property rights which grantsscientists temporary legal monopolies (generally patents) allowing producersto appropriate the returns In emphasizing a regulatory solution to theknowledge-production market failure the old economics of science literaturedid not attend to the question of how science plusmn several centuries old plusmnhistorically managed to address the public-goods problem In this sense theearly students in economics of science took the scientiregc rules norms andconventions as given to the extent they considered them at all (ibid 492)

Well before the advent of government science science evolved a set ofinstitutions to address its coordination problems25 The institution thataddresses the di culty of excluding non-payers once knowledge is madepublic is known as credit Scientists who produce a valuable idea receivepayment in the form of credit generally via citation and also via prizeseponymy and other forms of recognition In science reputation is the coin ofthe realm and reputation can be seen as the stock of previous credit (anddiscredit) (See Latour and Woolgar 1986)

If other scientists use your output (and acknowledge use) then you will berewarded with credit `Just as the market rewards knowledge which enablessomeone to o er goods and services which customers wish to acquire so thereputational system rewards those who produce new ideas which others can


put to use and if the goods or ideas are unwanted or defective they will beignored or criticizedrsquo (Loasby 1989 39)

To argue that scientists seek credit does not require them to be indi erentto truth Scientists likely want both This may be so intrinsically scientistswant to be right perhaps out of native curiosity even as they also want creditplusmn cashable in the form of higher wages collegial esteem promotions etc plusmn forbeing right Moreover cognitive and non-cognitive goals need not be at oddsplusmn a more veritistic result can (and often does) lead to for example greatercredit or greater proregt26

The evidence that scientists care about credit is compelling Examiningeconomistsrsquo salaries Hamermesh regnds that higher pay is robustly associatedwith greater citation (1989 cited in Colander 1989) Second there is therelative paucity of anonymous publication A credit-indi erent scientistmight well publish anonymously (analogous to the anonymous charitabledonor) but anonymously published results are the rare exception in thehistory of science27 Third consider the scientiregc institution of priority thescientiregc convention of awarding regrst discoverers all the credit Thecredit-indi erent scientist who has only cognitive goals should be indi erentto priority in discovery28

Because credit ordinarily comes only with priority the history of science isrife with priority disputes Robert Merton documents that the battles overpriority are regerce recurring and involve some of the greatest scientiregc namesNewton Hooke Liebniz Huygens Cavendish Watt Lavoisier FaradayLaPlace several Bernoullis Legendre and Gauss are some of the worthiesthat Merton identireges as having been involved in priority disputes (1973[1957] 286plusmn324) Itrsquos no di erent today The most eminent scientists stillsquabble over for example who regrst isolated the human immunoderegciencyvirus thought to cause AIDS (Hull 1997) Were scientists indi erent to creditwhich comes only with priority we would not observe the recurrent prioritydisputes so characteristic of scientiregc history

The way in which priority is determined helps answer our second questionwhy do scientists openly publish their results In academic science (1) creditis as noted generally awarded entirely to regrst discoverers ( priority) and (2)the regrst to publish is deemed the regrst discoverer It is unusual to get credit forresults not published or for results already published by others29 To getcredit you must be regrst to publish30 The tradeo for the scientist who wantscredit is clear waiting too long to publish risks losing priority and thereforecredit while rushing into print risks errors which if they ramify will lead todiscredit (Hull 1988 352) More pre-publication work increases reliabilityand thus expected credit but consumes time that reduces expected credit

Awarding credit based on priority and awarding priority based onpublication are rules that promote the virtues of more rapid scientiregcinnovation and broader dissemination of knowledge Merton referred tothe norm of open publication (and the idea that scientiregc knowledge is

156 Articles

common property) as an ethic of `communismrsquo (1973 [1942] 273plusmn5) Butinterested scientists can have incentives to keep their results secret or to delaypublication so the rules qua norms are not by themselves incentive com-patible The incentive to publish is created by the rule that assigns credit to theregrst to publish

What if proregt as well as credit is at stake The incentive to publish regrst canbe overwhelmed when the value of expected proregt which requires secrecyexceeds the value of expected credit Secrecy characterizes industrial sciencewhere proregtable ideas are made public generally only with patent protectionso that the owner of the intellectual property can appropriate returns viatemporary legal monopoly Dasgupta and David (1987) distinguish sciencefrom technology on this very basis plusmn science is they argue concerned withadding to the stock of scientiregc knowledge and thus has the practice of openpublication technology in contrast is concerned with maximizing rents froma given stock of knowledge and thus has a practice of secrecy31 If proregtsincrease at a faster rate than returns to academic reputation then one wouldexpect more defections from the academic model of open publication to thecommercial model of secrecy or an increased blurring of the line betweenacademic and commercial research32

Innovation and dissemination of ideas are not the only rationales foropenness so too is the policing function so characteristic of science plusmn thecertiregcation of ideas as reliable This takes us to our third query why scientistsproduce reliable knowledge There are two important scientiregc institutionshere peer review and replication

Like all agents individual scientists accept much on faith Were scientistsobliged to independently assay every bit of knowledge used in their researchscience would stop Reliability is nonetheless vital so science has evolved asocial system of trust which is built upon a process of veriregcation via peerreview and replication Scientists have some assurance that their knowledgeinputs are reliable when independent peer reviewers have directly assayed theideas in question or when subsequent ùsersrsquo if any have indirectly done so

Publication is generally necessary but not su cient for obtaining creditCredit accrues only when others use ideas and they prove reliable Ideas thatprove unreliable bring discredit and fall into disuse In the empirical sciencesthen there are strong incentives to produce reliable knowledge Researchoutput is `quality checkedrsquo not only by peer reviewers but also by subsequentusers

I do not wish to suggest that peer review and replication are withoutproblems Replication is not a simple matter practically or philosophically(Collins 1985) In some regelds economics included (Mayer 1993b) directpositive replications (in the sense of reproducing a published result) are rarelypublishable so the returns to direct replication are lower than the returns tooriginal research (Feigenbaum and Levy 1993)33 Some review processes aremore demanding than others the refereeing process is itself vulnerable to


corruption34 Moreover it is probably the fate of most published papers neverto be cited much less to be checked Holub et al (1991) regnd about half ofpeer-reviewed papers in economic growth theory are never cited and that 85percent are cited regve or fewer times35

For papers that survive publication replication is pervasive in theempirical sciences Those who use other scientistsrsquo results as an input verymuch want those results to be reliable Scientists are unlikely to sabotage theirown research by choosing inputs known to be faulty As such a kind ofindirect replication is in force because serious errors will ultimately manifestthemselves with adverse reputational consequences for the scientists who arethe source of the error (especially if due to fraud or carelessness)

Not all `publishedrsquo papers are peer reviewed Working papers andother papers not yet peer reviewed are increasingly used and cited byscientists Some repositories such as the Social Science Research Network(httpwwwssrncom) gather and index working papers Others such as theLos Alamos Physics Archive (httpxxxlanlgov) also allow for open peercommentary plusmn any reader can comment plusmn as against the traditional process ofrefereeing by editorially-designated experts Some journals augment peerreview with open peer commentary (On the virtues of open peer commentaryversus peer review see Harnad 2000)

Though all scientiregc claims are corrigible survival under the scrutiny ofuse is a tentative signal of reliability and allows potential users to rely on thejudgment of others Direct replication in contrast where reliability isindependently undertaken by the user is costly in time and expense Rivalryin science provides some incentive for scientists to undertake direct replica-tion It is unrealistic to expect scientists to rigorously attempt to refute theirown hypotheses But as Hull points out their rivals will be happy to do so(1988 4) Competition helps promote replication because competitors havean incentive to refute (or at least challenge) results they regnd inimical to theirown work

There is for example a well-known recent literature in labour economicswhich regnds that recent minimum-wage increases do not result in adverseemployment consequences for low-skilled workers (Card and Krueger 1995)Those who regnd this result congenial such as political proponents ofminimum-wage increases are unlikely to scrutinize Card and Kruegerrsquosmethods too closely But those who believe that a minimum wage doeshave adverse employment consequences and have built scientiregc reputationson this view are likely to examine Card and Kruegerrsquos controversial regndingsrather sceptically and have in fact done so36 When the stakes are highenough the ordinary disincentives to direct replications can be overcome

In economics where reliability is harder to come by than in the naturalsciences publication incentives with respect to positive replication can changeover time Goldfarb (1995) regnds in several empirical literatures that thereturns to positive replications (not checking per se but `conregrmingrsquo results)

158 Articles

are initially higher than negative replications (disconregrming results) but thatafter a period of normal science disconregrming results become relatively moreattractive that is likelier to be published (For more on replication ineconomics see Backhouse 1997 135plusmn57)

Because it helps promote replication rivalry in science works to helpeliminate error in a kind of self-correcting fashion Errors get exposed notbecause scientists disinterestedly refute their own pet theories but becausetheir interested rivals have partisan incentives to do so Moreover interestedrivals are generally the most qualireged reviewers Those with the greatestincentive to criticize are also those with the greatest expertise Competition forcredit induces scientists to rigorously assay their rivalsrsquo research and iswhat Popper referred to when he said `should individual scientists everbecome òbjective and rationalrsquo in the sense of ìmpartial and detachedrsquothen we should indeed regnd the revolutionary aspect of science barred by animpenetrable obstaclersquo (1975 93 cited in Hull 1988 359)

But competition among scientiregc rivals is tempered by a mutual depend-ency that arises from specialization in the division of cognitive labour andfrom the way in which credit is assigned Scientists produce and consumeknowledge or perhaps more accurately they use the output of otherscientists as a capital-good-type input in their own production (Ghiselin1989) As such scientists partly depend on suppliers who are also their rivalsin the output market Scientists are likely to beneregt cognitively from usingtheir rivalsrsquo results Rivals are after all working on similar problems andgiven specialization are the likeliest source for improving the reliability ofonersquos own work

Scientists need their competitorsrsquos work as an input and they also needtheir rivalsrsquo support A scientist wants other scientists to make use of hisoutput This is not merely professional pride but the fact that credit obtainsonly when fellow producers award it Scientists need their rivals they needtheir rivalsrsquo work to meet cognitive goals and need their rivalsrsquo attention toobtain the credit that comes only from use Thus is competition in sciencetempered by a kind of mutual dependency (on this see Hull 1998 1997)

Mutual dependency of rivals goes to another incentive question whyscientists generally give credit where credit is due Why for example doscientists generally practice honest citation when it is costless to plagiarize (inwhole or in part) and thereby receive credit for ideas one didnrsquot actuallyproduce One might get caught of course with the attendant reputationaldisaster likely to follow But plagiarism unlike unreliable work where errorseventually manifest in future research is not subject to indirect replicationThe determined plagiarist can go a surprisingly long time before detection(Kohn 1986)

Mutual interest among scientists however creates an incentive to citeothers in order to enlist their support which is necessary for obtaining creditWe need our fellow scientists in part for their output and in part for their


support without which credit for our own output cannot be had Honestcitation is not only a Mertonian norm plusmn respect for othersrsquo intellectualproperty plusmn nor is it motivated solely by fear of detection Mutual dependencealso makes it in the scientistrsquos interest to cite properly

6 CONCLUSION

The forgoing assumes that learned inquirers and everyday agents value andcan obtain reliable knowledge at reasonable cost which in turn presupposesan enterprise with the incentive and the ability to inquire empirically In someenterprises such as those that consider ethics or aesthetics or spiritualityreliable knowledge may be impossible In other places reliable knowledgemay be too costly or may not be highly valued In these domains where theway the world is cannot or does not meaningfully constrain what is believed tobe reliable knowledge is beside the point and the scientiregc rules I have arguedhelp promote reliable knowledge cannot do so

When science does succeed in proceeding reliable knowledge it succeeds inpart because its institutions are robust plusmn credit priority open publicationpeer review and replication The reputational system induces scientists toopenly and rapidly produce ideas by rewarding them with credit Peer reviewand (direct and indirect) replication provide additional incentives to produceideas that are reliable When scientists change their mind it is not merely amatter of personal virtue In the long run scientists have powerful incentivesto use reliable ideas and to eschew unreliable ones even when there is anunhappy di erence what they want to be true and what they take to be true(Galison 1987)

I do not claim and know no one who does that science always producesreliable knowledge Scientiregc rules norms and conventions are imperfectthey are not all self-enforcing and they likely remacrect path-dependantine ciencies But this is true of markets too Like markets science issuccessful because it has robust institutions and thus fails only some of thetime

ACKNOWLEDGMENTS

I gratefully acknowledge without implicating them in the much revisedarguments presented here the constructive criticisms of Wade Hands JimWible John Davis Alan Walstad Bob Goldfarb numerous seminar parti-cipants and the referees

Thomas C LeonardPrinceton University

tleonardprincetonedu

160 Articles

NOTES

1 The term `critical science studiesrsquo was suggested by a referee2 Some students of science take seriously the idea that scientiregc success is pure

happenstance Mary Hesse for example says `science might after all be amiraclersquo (1980 154)

3 Economics entered the science-studies derby late and the earliest retailers ofeconomic ideas were typically philosophers (Hull 1988 Rescher 1989 Kitcher1993) though there are exceptions such as Tullock 1966 Science studiesincumbents tend to be hostile or indi erent to economic thought at least itsmainstream variant (See Mirowski and Sent (2002) for a stimulatingcompendium of critiques) Even in economic methodology economicallyinformed approaches to methodology are comparatively uncommon Someexceptions again are Wible (1998) Colander (1989) Feigenbaum and Levy(1993) Mayer (1993a) For more on economics in science studies see Hands2001 chapter 8

4 Theories are regarded as empirically equivalent when they have identical empiri-cal consequences (Laudan and Leplin 1991 451) Since it is easy to generatetheoretical àlternativesrsquoby trivially changing a current theory (Cross 1998) thereare logically at least an inderegnite number of empirically equivalent theories Aninderegnite number of rival theories entailing the same empirical consequencesdoesseem to imply underdetermination since any actual evidence would support (ordisconregrm) all rivals equally (This paragraph and footnote is taken somewhataltered from Goldfarb et al 2001)

5 `Strategiesrsquo is plural by design a recognition that empirical methods in science areheterogeneousIan Hacking (1983 152) says `There is not just one way to build ahouse or even to grow tomatoes We shouldnot expect somethingas motley as thegrowth of knowledge to be strapped to one methodologyrsquo

6 The term `naturalismrsquo as used in the epistemological context is due to Quine(1969)

7 Thus notwithstanding his emphasis on scientiregc motivation Popper remainsclose to the traditional philosophers of science with respect to the nature andstatus of meta-scientiregc claims

8 Charles Sanders Peirce proposed it nearly a century earlier saying `philosophyought to imitate the successful sciences in its methods rsquo (cited in Wimsatt 1981124)

9 Cole (1992) from which the Merton citations are obtained reports that Mertonadded this quote as a note in the 1957 reprint of Mertonrsquos famous 1942 paper onscientiregc norms Intriguinglywhen the paper was added to the 1973 collection ofMertonrsquos papers Merton altered the last sentence to read `Sooner or latercompeting claims to validity are settled by universalistic criteriarsquo (Cole 1992 4)

10 Quinersquos (1953) read empirical equivalence of rival theories as entailing epistemo-logical equality thereby reducing theory choice to wholly non-evidential (his termis `pragmaticrsquo) considerationsBut it is far from clear that actual theoretical rivalsare empirically equivalent It may be that trivial changes donrsquot create genuinelydistinct theories which is why it is di cult to regnd real-world examples ofgenuinely distinct empirical equivalents and the few we have o er little warrantfor the sweeping claim that there are likely empirical equivalents to most theoriesQuine himself later suggested that incompatible empirically equivalent theoriesmight not be genuine rivals but instead verbal variants of the same theory(Quine 1981 23plusmn30 cited in Haack 1998 101 n 6) Laudan and Leplin argue thatunderdeterminationdoes not follow from empirical equivalence unless empirical


consequences are improperly identireged with evidential support When evidentialsupport for a theory is not limited to instances that are the direct consequences ofthe theory epistemologically warranted choices among rivals are still possibleOne way is indirect evidential support Consider two empirically equivalenthypotheses (H1 and H2) where only H1 is derivable from a more general theoryT which also implies another hypothesis H Empirical consequence (E) of H isobserved repeatedly (many instances) which supports H and thereby T alsolending indirect evidential support for H1 but not H2 E is not a consequenceof H1

or H2 but nonetheless o ers an evidential basis for preferring H1 ceteris paribus(Laudan and Leplin 1991 464) (This footnote is taken somewhat altered fromGoldfarb et al 2001)

11 CSS thus turns the Received Viewrsquos conception of theory justiregcation on its headThe Received View says that scientists accept a claim because there is empiricalwarrant for it Warrant (eventually)causes acceptanceCSS inverts the traditionalcausality it says acceptance is what determines warrant In some CSS versionswarrant for a claim is said to consist of acceptance Say David Bloor

Instead of deregning it as true belief knowledge for the sociologist is whatevermen take to be knowledge Of course knowledge must be distinguishedfrommere belief This can be done by reserving the term `knowledgersquo for what iscollectively endorsed leaving the individual and idiosyncratic to count as merebelief

(1976 2plusmn3 cited in MaEgrave ki 1992 62)

12 Philip Kitcher argues `To see how bizarre this [CSS view of how empiricalevidence is immaterial] is we should note that the point also seems to show thatsociety can have no bearing on what scientists acceptrsquo (1998 40 emphasis added)

13 Latour and Woolgar (1986) for example use ethnographic research methods instudyingwhat scientistsdo in the lab If their observationaldata were not intendedas evidence for the validity of their claims about laboratory life what were theyfor

14 It is a measure of the polarization in contemporary science studies that one feelsobliged to note what should go without saying there is nothing in a sociologicaloranthropological or historical approach to science that per se requires a radicallyskeptical attitude towards scientiregc knowledge And though again it should gowithout saying many science studies scholars do not endorse the CSS view

15 Philosophers have located counterfactual examples justireged true beliefs thatshouldnrsquot be regarded as knowledge (Gettier 1963)

16 Veritistic and economic value should not be confused Some epistemologicalpositions however such as the pragmatism of William James propose that truthconsists in something we regnd it valuable to believe

17 The term `spontaneous orderrsquo is due to Friedrich Hayek who meant it to bedistinguished from a consciously planned or designed order Hayek often citedAdam Ferguson in this respect who characterized spontaneously evolvedorders as those that àre of human action but not of human designrsquo(Hayek 1973 20)

18 The process is called invisible because it is ordinarily thought to work without theknowledge of the participants Whether this is true or a requirement of thederegnition is debated See Ullmann-Margalit 1978

19 Some writers reserve the term `the invisible backhandrsquo for such outcomes(Brennan and Petit 1993) Following convention I use ìnvisible handrsquo to referto beneregcial outcomes only plusmn the successful alignment of individual and collectiveinterests

162 Articles

20 The invisible-hand theorists do represent a profound change in political philo-sophy a shift from virtue to justice No longer is governmentrsquospurpose to morallyinstruct plusmn to attempt to create a virtuous citizenry plusmn it is rather to underwrite andenforce rules that are the essential preconditions to successful commercial societyNonetheless the invisible hand theorists were themselves in some sensemotivated by virtue They believed that in a commercial society their moremodern conception of government would better serve the common good Thesepoints are due to Jerry Muller (1993)

21 With no missing markets this is su cient to ensure ParetooptimalityCollusion isdesirable to agents but precludedby the (known) fact that no sustainablecoalitionof sellers (or buyers) can control a large enough share of the market to a ect price

22 This may help explain why Kitcherrsquos critics in economics insist on the labelìnvisible-handrsquo (eg Mirowski 1996) when Kitcher himself in a long and closelyreasoned book never once uses the term

23 My own view is that Goldman and Cox (1996) too quickly identify the casual`marketplace of ideasrsquo metaphor with neoclassical perfect competition and regndthat the latter will not produce more truth as claimed by proponentsof former Ina later revised version of Goldman and Cox (1996) Goldman more carefullydistinguishes the neoclassical economic sense of competitive markets plusmn pricerivalry to serve consumers plusmn from the marketplace-of-ideas sense which refersmore to open vigorous debate of diverse opinions (1999 189plusmn217)

24 I use `reliable knowledgersquo as a rough synonym for veritistically valuable beliefsthough the attributes that make a belief or practice veritistically valuable gobeyond reliability Wade Hands (2001 146plusmn148) notes that Goldman (1987)invokes other standards power (problem-solving ability) fecundity speed (howquickly beliefs are processed) and e ciency and observes that scientists may needto trade o these virtues when they are opposed

25 It is with reluctance that I use the term ìnstitutionrsquoas a rubric for rules norms andconventions (and their enforcement) in science The term ìnstitutionrsquo meansdi erent things to Old Institutionalists and New Institutionalists to say nothingof others and it is notoriously resistant to precise deregnition John R Commonsnoted the di culty of a decent deregnition in Commons 1934 Geo Hodgsonrsquos(1998) attempt to characterize the `hard core propositionsrsquoof old Institutionalismdoes not hazard an attempt at deregnition until 13 pages into the article providingan illustration of how the term remains elusive even when in skillful andsympathetic hands

26 Of course if the returns to fraud for example are deemed relatively higher thenproregt seeking need not promote truth seeking (see Wible 1998 43plusmn60)

27 The mathematicians who wrote under the pseudonym Nicholas Bourbaki are aninteresting exception as noted in Hull 1997

28 Even the pure truth-seeker however can see instrumental value in credit Ifcredit leads to more funding and to greater opportunities to produce know-ledge then credit can be an instrumental if not an intrinsic goal for the puretruth seeker

29 By `publishedrsquo I mean `made publicrsquo or òpenly disclosedrsquo Working papers orwebbed papers are in this sense published Publication in refereed outlets entailspeer review discussed below One can regnd historical examples of credit awardedto scientists who were second in a race or who never bothered to formally publishtheir results but they are exceptional Multiple independent discoveries such asthe invention of the calculus by both Newton and Liebniz are a di erent matter(see Merton 1973 343plusmn70)

30 Though see Steve Stiglerrsquos (1980) Law of Eponymy `No scientiregc discovery is


named after its original discoverrsquo If Stiglerrsquos Law self-applies then giving Stiglercredit may be giving credit where credit is not due

31 There are other incentive e ects which we do not take up here For examplepriorityrsquos winner-take-all structure creates incentives for rapid innovation but italso can result in wasteful ex post gamesmanship (Merton 1973 317) and insocially ine cient expendituresanalogousto the patent-raceproblem(Daspguptaand David 1994)

32 The academicnorm of opennessand the commercialnorm of secrecy conmacrict in aninteresting way when academic research threatens intellectual property protec-tion Academic computer scientists for examplewho want to publish papers thatshow how to circumvent the copy-protection codes in software and recordedmusic often face litigation by industry groups often even when regrms hire them toregnd such weaknesses

33 Replication is rare in economics and econometric results that are revisited arenotoriouslyhard to reproduceAn organizedattemptat replication fundedby theUS National Science Foundation reviewed empirical papers submitted to theJournal of Money Credit and Banking from 1980 to 1982 (DeWald et al 1986) Of154 original authors notireged only in 90 cases were authors willing and able tosupply data and programs The replicators reviewed 54 of these data sets andfound that only eight were su ciently free of problems to permit an attempt atreplication And only in two of the remaining eight papers were the resultsactually reproduced in full The review team was thus able to fully reproduceeconometric results only 37 per cent (254) of the time Replicators wereattempting replication only in the narrow sense of reproducing results usingoriginal data sets and statistical procedures

34 Arthur Diamond (1986) estimates the present value of an additional publication(in 1994 dollars) to a 35-year-old academic mathematician as about $6750 (citedin Stephan 1996 1203) The editor of a journal which publishes 50 articles a yearcould develop a remunerativesideline particularlywhere the marginaldi erencesamong published and unpublished papers are small Less egregious and morecommon forms of corruption are nepotism and logrolling

35 In a sample of the 1187 papers published in the 1963 volume of Physical Reviewone half received zero or one citations in the subsequent three years and only15 per cent received six or more citations (Cole 1992 16)

36 See for example Neumark and Wascher (1995) and Hamermesh (1995) Formore on the controversy see Leonard (2000)

REFERENCES

ArchibaldG (1967) `Refutationor comparisonrsquoBritish Journal for the PhilosophyofScience 17 279plusmn96

Aronowitz Stanley (1988) Science as Power Discourse and Ideology in ModernSociety Minneapolis University of Minnesota Press

Arrow Kenneth (1962) Èconomic welfare and the allocation of resources forunventionrsquo in RR Nelson (ed) The Rate and Direction of Inventive ActivityEconomicand Social Factors PrincetonNJ Princeton University Press pp 609plusmn25

Backhouse Roger (1997) Truth and Progress in Economic Knowledge CheltenhamUK Edward Elgar

Brennan Geo rey and Pettit Philip (1993) `Hands invisible and intangiblersquo Synthese94 191plusmn225

Callebaut Werner (1993) Taking the Naturalist Turn or How Real Philosophy ofScience is Done Chicago University of Chicago Press

164 Articles

Card Davidand KruegerAlan (1995)Myth and MeasurementThe New EconomicsofThe Minimum Wage Princeton NJ Princeton University Press

Colander David (1989) `Research on the economics professionrsquo Journal of EconomicPerspectives 3(4) 137plusmn48

Cole Stephen (1992) Making Science Between Nature and Society Cambridge MAHarvard University Press

Collins Harry (1981) `Stages in the empirical programmeof relativismrsquoSocial Studiesof Science 11 3plusmn10

Collins Harry (1985)ChangingOrderReplicationand Induction in Scientiregc PracticesLondon Sage Publications

Collins Harry and Yearly Steven (1992) `Journey into spacersquo in A Pickering(ed) Science as Practice and Culture Chicago University of Chicago Press369plusmn89

Commons John R (1934) Institutional Economics Its Place in Political EconomymNew York Macmillan

Cross Rod (1998) `The DuhemplusmnQuine hypothesisrsquo in John B Davis D Wade Handsand Ischial MaEgrave ki (eds) The Handbookof Economic Methodology Cheltenham UKEdward Elgar pp 107plusmn10

DaspguptaPartha and David Paul (1987) Ìnformationdisclosureand the economicsof science and technologyrsquo In George Feiwel (ed) Arrow and the Ascent of ModernEconomic Theory London Macmillan Press pp 519plusmn42

Daspgupta Partha and David Paul (1994) `Towards a new economics of sciencersquoResearch Policy 23 487plusmn521

Demsetz Harold (1993) `The theory of the regrm revistedrsquo in Oliver Williamson andSidney Winter (eds) The Nature of The Firm New York Oxford University Presspp 159plusmn78

Davis John B (1997) `The fox and the henhouses the economics of scientiregcknowledgersquo History of Political Economy 29(4) 741plusmn46

DeWald WG Thursby JG and Anderson RG (1986) `Replication in empiricaleconomics the Journal of Money Credit and Banking projectrsquo American EconomicReview 76(4) 587plusmn603

DiamondArthur (1986) `How much is a citationworthrsquo Journalof Human Resources21(2) 200plusmn15

Feigenbaum Susan and Levy David (1993) `The market for (Ir)reproducibleeconometricsrsquo Social Epistemology 7(3) 215plusmn32

Feyerabend Paul (1975) Against Method Outline of an Anarchistic Theory ofKnowledge London Verso

Galison Peter (1987) How Experiments End Chicago University of Chicago PressGettier Edmund (1963) Ìs justireged true belief knowledgersquo Analysis 23 121plusmn23Ghiselin Michael (1989) Intellectual Compromise New York Paragon HouseGoldfarb Robert (1995) `The economist-as-audience needs a methodology of

plausible inferencersquo Journal of Economic Methodology 2(2) 201plusmn22Goldfarb Robert Leonard Thomas C and Suranovic Steven (2001) Àre rival

theories of smoking underdeterminedrsquo Journal of Economic Methodology 8(2)229plusmn51

Goldman Alvin (1986) Epistemology and Cognition Cambridge MA HarvardUniversity Press

Goldman Alvin (1987) `Foundations of social epistemicsrsquo Synthese 73 109plusmn44Goldman Alvin (1995a) `Reliablismrsquo in Robert Audi (ed) The Cambridge Dictionary

of Philosophy Cambridge University Press 693Goldman Alvin (1995b) `Social epistemologyrsquo in Robert Audi (ed) The Cambridge

Dictionary of Philosophy Cambridge University Press 746


Goldman Alvin (1999) Knowledge in a Social World Cambridge CambridgeUniversity Press

Goldman Alvin and Cox James (1996) `Speech truth and the free market for ideasrsquoLegal Theory 2 1plusmn32

Goldman Alvin and Shaked Moishe (1991) Àn economic model of scientiregc activityand truth acquisitionrsquo Philosophical Studies 63 31plusmn55

Haack Susan (1998) Manifesto of a Passionate Moderate Chicago University ofChicago Press

Hacking Ian (1983) Representing and Intervening Cambridge Cambridge UniversityPress

Hamermesh Daniel (1995) `Commentrsquo for Review Symposium on Myth andMeasurement The New Economics of the Minimum Wage Industrial and LaborRelations Review 48(4) 835plusmn38

HandsD Wade (1995) `Social Epistemologymeets the Invisible Hand Kitcher on theAdvancement of Sciencersquo Dialogue 34 605plusmn21

Hands D Wade (1997) `Caveat emptor economics and contemporary philosophy ofsciencersquo Philosophy of Science (Proceedings) 64 S107plusmn16

Hands D Wade (2001) Remacrection Without Rules Economic Methodology andContemporary Science Theory Cambridge Cambridge University Press

Harnad Steven (2000) `The invisible hand of peer reviewrsquo Exploit Interactive 5(April)

Hausman Daniel (1992) Essays on Philosophy and Economic MethodologyCambridge Cambridge University Press

Hayek Friedrich (1973) Law Legislation and Liberty Volume One Rules and OrderChicago University of Chicago Press

Hesse Mary (1980) Revolutions and Reconstructions in the Philosophy of ScienceBloomington IN Indiana University Press

Hodgson Geo (1998) `The approach of institutional economicsrsquo Journal ofEconomic Literature 36 166plusmn92

Hollis Martin (1994) The Philosophy of Social Science Cambridge CambridgeUniversity Press

Holub Hans Werner TappeinerGottfried and EberharterVeronika (1991) `The ironlaw of important articlesrsquo Southern Economic Journal (58) 317plusmn28

Hull David (1988) Science as a Process Chicago University of Chicago PressHull David (1997) `Whatrsquos wrong with invisible-hand explanationsrsquo Philosophy of

Science (Proceedings) 64(4) S117plusmn26Kitcher Philip (1993) The Advancement of Science Science Without Legend

Objectivity Without Illusions Oxford Oxford University PressKitcher Philip (1998) À plea for science studiesrsquo in Noretta Koertge (ed) A House

Built on Sand New York Oxford University Press pp 32plusmn56Kohn Alexander (1986) False Prophets Oxford Basil BlackwellKuhn Thomas (1992) The Trouble with the Historical Philosophy of Science (Robert

and Maurine Rothschild Distinguished Lecture) Cambridge Department of theHistory of Science Harvard University

Kuhn Thomas (1996) [1962] The Structure of Scientiregc Revolutions 3rd editionChicago University of Chicago Press

Lakatos Imre (1978) The Methodology of Scientiregc Research Programs Volume IJohn Worrall and Gregory Currie (eds) Cambridge Cambridge University Press

Latour Bruno (1987) Science in Action Cambridge MA Harvard UniversityPress

Latour Bruno and Steve Woolgar (1986) [1979] Laboratory Life The Construction ofScientiregc Facts 2nd edition Princeton NJ Princeton University Press

166 Articles

Laudan Larry (1984) Science and Values The Aims of Science and Their Role inScientiregc Debate Berkeley University of California Press

Laudan Larry (1990) Science and Relativism Some Key Controversies in thePhilosophy of Science Chicago University of Chicago Press

Laudan Larry and Leplin Jarrett (1991) Èmpirical Equivalence and Under-determinationrsquo The Journal of Philosophy 88(9) 449plusmn72

Leonard Thomas (2000) `The very idea of applied economics the modern minimum-wage controversyand its antecedentsrsquoHistory of Political Economy Supplement toVolume 32 Roger Backhouse and Je Biddle (eds) Durham NC Duke UniversityPress 117plusmn44

Loasby Brian (1989) The Mind and Method of the Economist Brookregeld VT EdwardElgar

MaEgrave ki Uskali (1992) `Social conditioning of economicsrsquo in Neil de Marchi (ed)Post-Popperian Methodology of Economics Recovering Practice Boston Kluwer65plusmn104

Mayer Thomas (1993a) Truth versus Precision in Economics Brookregeld VT EdwardElgar

Mayer Thomas (1993b) `Commentary on The Scientiregc Status of EconometricsrsquoSocial Epistemology 7(3) 269plusmn73

Merton Robert (1973) The Sociology of Science Norman Storer (ed) ChicagoUniversity of Chicago Press

Mirowski Philip (1996) `The economic consequences of Philip Kitcherrsquo SocialEpistemology 10 153plusmn170

Mirowski Philip and Sent Esther-Mirjam (eds) (2002) Science Bought and Sold TheNew Economics of Science Chicago University of Chicago Press

Muller Jerry (1993) Adam Smith in His Time and in Ours Designing the DecentSociety New York Free Press

Nelson Richard (1959) `The simple economics of basic scientiregc researchrsquo Journal ofPolitical Economy 67(3) 297plusmn306

Neumark David and Wascher William (1995) `The e ect of New Jerseyrsquos minimumwage increase on fast-food employment a reevaluation using payroll recordsrsquoNBER Working Paper No 5224 Cambridge MA NBER

Peirce Charles Sanders (1877) `The regxation of belief rsquo Population Science Monthly 121plusmn15

Popper Karl (1959) The Logic of Scientiregc Discovery London HutchisonPopper Karl (1972) Objective Knowledge An Evolutionary Approach Oxford

Clarendon PressPopper Karl (1975) `The rationality of scientiregc revolutionsrsquo in Rom HarreAcirc (ed)

Problems of Scientiregc Revolution Oxford Clarendon Press 72plusmn101Quine Willard (1953) `Two dogmas of empiricismrsquo in From A Logical Point of View

Cambridge MA Harvard University Press pp 20plusmn46Quine Willard (1969) Èpistemology naturalizedrsquo in Ontological Relativity and

Other Essays in the Philosophy of Science New York Columbia UniversityPress 69plusmn90

Rescher Nicholas (1989) Cognitive Economy The Economic Dimension of The Theoryof Knowledge Pittsburgh University of Pittsburgh Press

Rescher Nicholas (1993) Pluralism Against the Demand for Consensus OxfordClarendon Press

Smith Adam (1937) [1776] An Inquiry into The Nature and Causes of The Wealth ofNations New York Modern Library

Stephan Paula (1996) `The economics of sciencersquo Journal of Economic Literature34(3) 1199plusmn1235


Stigler Steve M (1980) `Stiglerrsquos law of eponymyrsquo Transactions of the New YorkAcademy of Sciences 39147plusmn57

Suppe Fred (ed) (1977) The Structure of Scientiregc Theories 2nd ed UrbanaUniversity of Illinois Press

Tullock Gordon(1966) The Organizationof Inquiry DurhamDuke UniversityPressVaughan Karen (1987) Ìnvisible handrsquo in John Eatwell Murray Milgate and Peter

Newman (eds) The New Palgrave The Invisible Hand New York WW Nortonpp 168plusmn72

Ullmann-Margalit Edna (1978) Ìnvisible Hand Explanationsrsquo Synthese 39(2)263plusmn91

Walstad Alan (2001) Òn science as a free marketrsquo ManuscriptWible James (1998) The Economics of Science Methodology and Epistemology as if

Economics Really Mattered London RoutledgeWimsatt William (1981) `Robustness reliability and overdeterminationrsquo in Marilyn

Brewer and Barry Collins (eds) Scientiregc Inquiry and the Social Sciences SanFrancisco Jossey-Bass pp 124plusmn63

168 Articles

was to reduce that success to a set of logical even algorithmic rules forgrinding out truth claims sometimes known as Scientiregc Method But theReceived View failed to o er any theory of scientiregc motivation and thuswhatever the merits of its normative case for Method had no way of knowingwhether real scientists had the right incentives to follow the Scientiregc MethodWhat I call Critical Science Studies (CSS) su ers from the obverse problem1

It o ers a theory of scientiregc motivation portraying scientists as (at leastpartly) self-interested and as having worldly goals similar to economic agentsAt the same time however CSS abandons as impossible the entire normativeproject of the theory of science CSS scholars defend and often assume`cognitive egalitarianismrsquo `the thesis that all beliefs are epistemically orevidentially equal in terms of their supportrsquo (Laudan 1984 29plusmn31) TheReceived View has a theory of scientiregc success but no theory of scientiregcmotivation CSS has a theory of scientiregc motivation but denies any prospectfor (epistemologically meaningful) scientiregc success Thus neither can answerthe important question plusmn can successful science accommodate a realistic viewof scientiregc motivation plusmn because both regard the question as immaterial

This paperrsquos premise is that the question deserves an answer that is that anadequate theory of science must o er both a theory of scientiregc motivationand a theory of scientiregc success I will not argue for this premise except asfollows We need a theory of scientiregc success because the natural scienceswhile fallible and imperfect and not immune to politics bias fashion and fadhave nonetheless been the most successful of human cognitive endeavours(Haack 1998 130) Unless sciencersquos success is denied or can be regarded as anaccident of happenstance or as inherently inexplicable itrsquos worth investigat-ing how it is accomplished2 The idea is that science studies must not onlydescribe the practices of scientists but analyse how scientiregc practices succeedor fail in accomplishing cognitive goals

We also need a theory of scientiregc motivation even if it is the theory thatscientists plusmn professional skeptics plusmn will blindly heed the rules laid down forthem by lawgiving philosophers of science If one regards scientists assometimes having interests opposed to what the rules dictate then it matterswhether the rules are incentive compatible The history of science suggeststhat the rules are frequently not observed There may be something wrongwith the rules (Feyerabendrsquos 1975 conclusion for example) or alternativelythe rules may be regne but there are incentive problems or both Without atheory of scientiregc motivation there is no basis for distinguishing amongthese rival explanations and in addition there is no prospect for identifyingscientiregcally good rules that interested scientists will opt to follow

The present task is to ask under what circumstances can a real communityof fallible èpistemically sulliedrsquo (Kitcher 1993 384) inquirers achievethe good outcomes traditionally thought to require ideal inquirers (Haack1998 98) Invisible-hand explanations suggest themselves naturally to aneconomist plusmn they are deeply rooted in the disciplinary ethos This paper o ers

142 Articles












144 Articles






(1978 4)








(ibid)


146 Articles














148 Articles












(1988 12plusmn13)



150 Articles


(Kuhn 1992 9)










(1937 [1776] 423)


152 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles












144 Articles






(1978 4)








(ibid)


146 Articles














148 Articles












(1988 12plusmn13)



150 Articles


(Kuhn 1992 9)










(1937 [1776] 423)


152 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles







144 Articles






(1978 4)








(ibid)


146 Articles














148 Articles












(1988 12plusmn13)



150 Articles


(Kuhn 1992 9)










(1937 [1776] 423)


152 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles






(1978 4)








(ibid)


146 Articles














148 Articles












(1988 12plusmn13)



150 Articles


(Kuhn 1992 9)










(1937 [1776] 423)


152 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles






(ibid)


146 Articles














148 Articles












(1988 12plusmn13)



150 Articles


(Kuhn 1992 9)










(1937 [1776] 423)


152 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles














148 Articles












(1988 12plusmn13)



150 Articles


(Kuhn 1992 9)










(1937 [1776] 423)


152 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles






148 Articles












(1988 12plusmn13)



150 Articles


(Kuhn 1992 9)










(1937 [1776] 423)


152 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles












(1988 12plusmn13)



150 Articles


(Kuhn 1992 9)










(1937 [1776] 423)


152 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles







(1988 12plusmn13)



150 Articles


(Kuhn 1992 9)










(1937 [1776] 423)


152 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles


(Kuhn 1992 9)










(1937 [1776] 423)


152 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles






(1937 [1776] 423)


152 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles











154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles






154 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles















156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles







156 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles














158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles







158 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles









6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles


6 CONCLUSION




ACKNOWLEDGMENTS




160 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles

NOTES

























162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles














162 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles




















REFERENCES







164 Articles


















































166 Articles


































168 Articles








REFERENCES







164 Articles


















































166 Articles


































168 Articles


















































166 Articles


































168 Articles


























166 Articles


































168 Articles


































168 Articles









168 Articles

Reflection on rules in science: an invisible-hand …tleonard/papers/Rules_in_science.pdf · Re¯ ection on rules in science: an invisible-hand perspective Thomas C. Leonard ... Imre

Documents