of and "The" Scientific Jim Griesemer, an assistant of philosophy at the Univ. of Califomia, Davis, CA 95616, did his undergraduatework in ge- netics at the Univ. of Califomia at Berkeley. There he became in- Method terested in the foundations evolutionary , going on to a master's in and a Ph.D. in the Committee on Conceptual Foundations of Science of the Univ. of Chicago. He has worked on the development of exhibits at the of Science and In- dustry in Chicago, and is currently doing researchin the founda- tions of multi-level selection theory, representations of scientific Jim Griesemer reasoning, and the institutional of biology. Downloaded from http://online.ucpress.edu/abt/article-pdf/47/4/211/86196/4448021.pdf by guest on 01 October 2021

The od be fundamentally changed, and indeed, the that there ever was a definitive addresses two central ques- conception of is a stem- tions: (1) what is the structure of a ?, ming from an ahistorical, nondevelopmental and (2) what are the processes of growth or change view of philosophy of science. of science over ? The gulf between philosophical 3. One should therefore be circumspect in teach- treatments of these two questions is one of the great ing "the" scientific method or wielding "the" ironies of a discipline which prides itself on system- scientific method in argument against propo- aticity, comprehensiveness, and coherence. Separate nents of opposing scientific views. traditions have emerged since the 1960s to deal with There is a pernicious , or false co- the two questions, rendering the gulf between them herence or robustness, about this myth of the scien- nearly institutionalized; the inheritors of logical em- tific method which stems from the that it has piricism, the formalists, analyze the structure of the- two rather different sources of support. One source ories in formal terms which promote an ahistorical, is the traditional emphasis of science asocial of scientific change. The new histor- have placed on formalist accounts of theory struc- ical movement, heralded by Kuhn, but prefigured ture. The logician's deductive standard for scientific by others, explains the of growth and and , and the attendant view change in science with socio-historicalmodels of sci- of as sets of and rules in a formal entific , leaving the structure of the products of , leads to the that there is only one pos- science unanalyzed entirely or only inchoately spec- sible scientific method because there is only one ified and incommensurable with the analyses of the standard of logical deduction to all and only valid formalists. conclusions [see Suppe (1977) for a detailed account As this gap has become recognized as the sort of of the history of these ; Nagel (1961) and foundational problem which can lead either to the Hempel (1965) give important statements of the subdivision of the into a formalist program and view]. "The" scientific method is, accordingly, the a socio-historicalprogram, or to a grand unification, testing of universal statements offered as scientific philosophy of science has entered a period of tumult by comparing deductive consequences of laws and transition. In this essay I wish to draw a lesson with statements of observational about the from these which has not been widely world. (1965) calls it the method of appreciated outside history, philosophy, and so- "conjecturesand refutations." ciology of science that has particularrelevance to sci- The other main source of the myth of the scientific ence educators. The argument I wish to make is as method comes from a historical view of science, con- follows: ditioned by philosophical treatments (by both phi- 1. A clearer conception of scientific method is rec- losophers and ), which fails to recognize ognized on both sides as a central element of any of goals, interests, or motives among any possible reconciliationbetween the two tra- the people who do science that might influence their ditions in philosophy of science. methods. This view is a species of "" 2. Regardless of the outcome (philosophy of sci- applied to science, i.e., that there is a timeless es- ence split into two or more disciplines or grand sence of science which captures the important fea- unification), our conception of scientific meth- tures, all others mere distraction [see Hull

PHILOSOPHY OF SCIENCE 211 (1975, 1983) for extensive criticism of essentialism in The FormalistConception of history and philosophy of science]. Science seeks Explanationand ScientificChange "the" about the world, and the rational scien- The deductive-nomological model of explanation tist pursues "the" best method for attaining truth. (see Hempel 1965, 1966) epitomizes the best and This conception of science led Popper (1959, p. 15) worst of the formalist approach. are to argue that "The central problem of deductive , on this view, with the conclu- has always been and still is the problem of the sion being the "explanandum" and the premises growth of . And thegrowth of knowledgecan consisting of sentences stating general laws and be studiedbest by studyingthe growthof scientificknowl- spe- cial rules (see Hempel 1966, Ch. 5 for a brief intro- edge."This presumably follows from the assumption duction). The explanandum is that if anyone gains knowledge, it's the scientists, so typically supposed to describe an or of The we should study them in order to develop a para- affairs. chief of this "covering- model" (where the digm of the . laws deductively "cover" the explanandum) is the The turmoil in philosophy of science today stems, rigor with which deductive can be made. in part, from a away from essentialist thinking Only true conclusions can follow from (Hull 1975, 1983; Toulmin 1967, 1971, 1972) which deductively true premises, so it appears very powerful to consid- has, in turn, been motivated by increased considera- Downloaded from http://online.ucpress.edu/abt/article-pdf/47/4/211/86196/4448021.pdf by guest on 01 October 2021 er as a form of deductive . tion of biology as a subject of philosophical study. explanation Several sorts of when Biology is much less easily pigeon-holed by formalist difficulty arise, however, this analysis is applied to scientific practice. First, approaches than has been in the . Biolo- of formal gists would be hard pressed to find much biological the technical difficulties analysis of scien- tific theories expressed in natural insight in Woodger's (1937) attempt to axiomatize bi- language are for- midable. A comparison of the text of Darwin's ology, though Mary Williams' efforts (1970) are of (1859) On the Origin with more interest precisely because they show how far a of Species Mary Williams' (1970) system sobers one to the reform of our of deduction is required to re- immensity of the task. since theories are main within the outlines of formalism. On the one , viewed as linguistic structures sets of hand biology is much less highly mathematized than (i.e., sentences), deep problems of epistemology arise in linking scientific theories physics, so it is more difficult to see how to proceed to the world. sentences can be from ax- with formal analysis. On the other hand, much of Only derived what are the criteria the existing historical scholarship on biology has de- ioms, so by which we judge that a particularexplanandum veloped in the same period as the philosophical de- sentence corresponds to an event or state of affairs in the velopments under discussion, so there is a reduced world? is on this word-world relation. tendency toward the kind of linear story-telling silent of which has added to the false sense of correctness of Third, the singular goal deductive inference, to that truth will essentialist analyses of physics. guarantee only emerge from scientific In the remainder of this essay I will discuss se- arguments with true premises, elevates truth and to an lected features of the two traditions, draw some im- explanation excessively exalted height. Deduc- tive to plications for our of scientific meth- logic provides no machinery analyze the crea- tion, or dissemination, or reception of scientific od, and describe some new directions for ideas. Popper (1959, p. 31) addresses this which emphasize the importance of the pedagogical problem a hard line limitations of "the" scientific method. Because phi- by drawing between the process of scien- tific on the one hand and losophy of science research is (1) more the process of scientific justification on the other, delegating the pluralistic in its conception of method and simul- former process to the taneously more cautious in generalizing from one special and retaining only the hard nut with some for domain of science to others, and (2) more centrally promise deductive in his He concerned to base answers to bothtraditional - analysis program. writes, sophical questions about science on a broad account The initial stage, the of conceiving or inventing a of scientific method, educators should discuss ques- theory, seems to me neither to call for logical analysis nor to be susceptible of it. The question how it hap- tions of method with historical and philosophical pens that a new idea occurs to a man-whether it is a sensitivity. Students who learn that "the" scientific musical theme, a dramaticconflict, or a scientific the- method is a myth will be less inclined to view sci- ory-may be of great interest to empirical psychol- ence as a cookbook procedure and more able to re- ogy; but it is irrelevant to the logical analysis of scien- flect on the and excitement of scientific tificknowledge. practice. As the Fourth, when real scientists (as opposed to ideally might put it, science is a way of worldmaking, and rational ones who never make mistakes, have an in- students need the proper tools to reach this larger finite amount of time to make all the necessary de- vision. ductions, etc.) are found in violation of the deduc-

212 THE AMERICAN BIOLOGY TEACHER, VOLUME 47, NO. 4, APRIL 1985 tive standard, the formalist has no alternative than the hallmarks of political revolutions, though for the to judge the mistaken or irrational.There is most part science conformed to the formalist . no obvious way to challenge the descriptive ade- Kuhn designated the latter "," and it quacy of the deductive analysis itself, because it differs from the formalist line in that it is governed takes no or regulatory force in virtue of by a installed by the revolution-deduc- the status of deductive logic as a formal rather than tive logic is not necessarily the only possible su- an empirical science. preme standard of rationality. Indeed, Kuhn's great This list of difficulties skims only a few features of accomplishment was to convince that they the important period in the history of philosophy of had to go out and look at cases in order to under- science from the 1920s to the 1960s (see Suppe 1977 stand how the standard of rationality itself can be and Brown 1977 for fuller historical treatments). The changed by . points I wish to draw attention to are: (1) that the This is not the place to recount either Kuhn's the- emphasis on logic pushed explanation to center ory-his book is eminently readable-nor to discuss stage and (2) that the of the formalist pro- the many criticisms of his views (see, e.g., Shapere gram (by successive weakening to the point where it 1964, Lakatos and Musgrave 1970 for some repre- retained little of its former grand claims), which was sentative and classic criticisms). However, it is

acknowledged even by some of its proponents, lent important to note that many of them focused on a Downloaded from http://online.ucpress.edu/abt/article-pdf/47/4/211/86196/4448021.pdf by guest on 01 October 2021 credence to the development of an alternativehistor- doctrine Kuhn called "incommensurability" which ical analysis of science. challenged directly the formalist . Kuhn con- The model of scientific stemming from sidered it a consequence of scientific revolution that the formalist view is perhaps the weakest part of a whole new way of "seeing" the world arises which this whole tradition. Since explanations are deduc- transforms language, meaning, and understanding. tive arguments, science progresses by finding more In effect, scientists on opposite sides of the revolu- general laws from which old laws can be derived as tionary divide cannot communicate their disagree- consequences and thereby explained. Science grows ments and they talk past one another. This is be- by incorporatingwhat we already know under more cause the standard of rationality itself is paradigm- fundamental, general laws which also permit new dependent and there is no higher court of appeal to deductions. All of science is a monumental deduc- resolve transparadigmdisagreement. tive explanation of on this view! Progress Many critics attacked the notion of incommen- is incremental, adding new layers, and explaining surability as incoherent, or untrue to the historical old science in the bargain. This has the undesirable facts (many scientists do seem to communicate effec- feature of never allowing us to analyze any state- tively across the divide), or as suggest- ment as formerly regarded as true, but now recog- ing that scientific change is irrational(since there are nized to be false, since on the deductive model of no standards for rationally changing ). explanation, nothing false can ever be explained. If The point I wish to make here is that the real failure past science was mistaken, the ball is passed to the of the doctrine of incommensurabilityis that it is not psychologists; if not, we have a logical explicit enough as a theory of formal relations be- on our hands. The only exception might be that the tween statements under different paradigms. Kuhn old can be considered to be a special case of a has given the historical impetus to look for an alter- more general law. Kepler's laws of planetary native to the formalist analysis of the language of have often been given as an example of laws which science, but has not provided a very good analytical follow, in the limit of large relative of the tool to do the looking. The formalists had very to the , as a special case of Newton's laws of powerful tools which seem to work only on remote motion and universal gravitation (see, e.g., Nickles islands of forever removed from the 1973 for a careful discussion). Moreover, the for- practiceof actual science. malist view recognizes explanation as the only his- torical process with which to ground an account of Model-BuildingStrategies scientific progress. The current challenge in philosophy of science, then, is twofold: (1) to develop new analytical tools The HistoricalApproach to which allow us to study the multifactedissues of sci- Explanationand ScientificChange entific method, including aspects of both discovery and justification, and (2) to develop models of scien- (1970) did a great deal to draw at- tific processes and test them with historical case tention to the importance of the insights of the histo- studies. I can only indicate here some current trends rian for the issues under consideration by in research, but it is important to recognize that a philosophers of science. In particular,Kuhn pointed much larger body of work is under development out that many historical developments bear some of which not only addresses this challenge, but is also

PHILOSOPHY OF SCIENCE 213 transformingthe Popperian image of scientific meth- the sake of ), but on the view, od as of a , deduction of con- multiple means of explanation, production, in- sequences, refutation, and so on. ference, etc. is an important component of justifica- An important first step has been to reexamine the tion (see Wimsatt 1981). Wimsatt (1980) discusses a requirements placed on scientists by the deductive of used in model-building strat- standard. Wimsatt (1980) has developed Simon's egies which link inference schemes with conditions (1957) conception of a heuristicprocedure into an al- under which a scientific theory must be altered. ternative standard for scientific reasoning. Wimsatt One difficulty of the science-as-problem-solving observed that the formalist analysis rests on numer- paradigm (see Laudan 1977 for an early formulation) ous in principleclaims about the powers of is that problem-solving appears to be too generic a problem-solversthat treat scientists, in effect, as uni- category to distinguish what is uniquely scientific versal Turing machines capable of solving any com- from other kinds of work. Hull's (1981, 1982) treat- putable problem (see Wimsatt 1980, section 3). But ment of the essentialism problem, however, sug- many easily specified problems are, even if in princi- gests that it may be a mistake to suppose science has ple soluble, practically unsolvable due to limitations a neat demarcation from other activities. The combi- of time, computational resources, , etc. nation of Wimsatt's heuristics theory and Hull's The point is that the deductive standard is fine for novel treatment of scientific fields as historical en- Downloaded from http://online.ucpress.edu/abt/article-pdf/47/4/211/86196/4448021.pdf by guest on 01 October 2021 an scientist, but not for a human being. Wim- tities of the same sort as biological species (with all satt (1980, p. 220) summarized Simon's insight as the same problems of as the latter!) sug- follows, gests a new way of looking at the problem of the Simonespoused a 'principleof boundedrationality' growth of knowledge. If scientists are heuristic prob- (Simon,1957, pp. 198-199)which assertsthat we are lem-solvers working in historical traditions (and generally faced with problems of such complexity pass on their methods and ideas in to how that we cannotsolve them exactly,and therefore,if biological pass on their genes), how do we areto get any solutionsat all we mustdo so by in- scientific theories evolve? troducing various simplifying and approximative techniques.Thus was bornthe idea of a heuristic. One inroad on this problem is to consider how sci- entists interact with one another, convey ideas, ar- Heuristic procedures, in contrast to algorithmic (de- gue with one another, come to agreement, negotiate ductive) ones, don't guarantee anything. They give controversy, etc. If we apply Wimsatt's conception answers which are good enough most of the time. of heuristic model-building strategies "one level up" The errors they produce tend to be systematic rather to the problems scientists face in communicatingsci- than random, and it is thus feasible to catalogue and ence, we can begin to build a structural picture of study errors in the application of heuristics. the steps involved in scientific change. To carry the There are two immediately important conse- biological analogy further, we could try to develop quences of this idea. An alternative to the deductive "transmission rules" which lead to the propagation standard is provided which appears promising as a of scientific ideas. This would involve careful study tool for analyzing scientific inferences which lead to of historical cases, since there is no more to explanations, thus addressing the formalist issue. suppose that scientists in different fields communi- Specifying the relation between heuristic procedures cate in the same way than there is to suppose that and the structure of theories is a still more difficult reproduction in biological species works the same problem, but the semantic view of theories (Suppes way. By developing the heuristics approach at the 1967; van Fraassen 1980; Beatty 1981; Thompson levels of individual problem-solvers, communicating 1983; Lloyd 1984; Griesemer 1984) provides a novel groups, and the , the two tradi- viewpoint. tional problems in philosophy of science would be The second consequence of the idea of heuristics coupled. The task then would be to dy- is that a path is opened up to consider problems of namic models of scientific activity at each level and theory structure and the growth of knowledge investigate the effects of events at one level on the jointly. First note that problem-solving rather than others. explanation is at the center of concern. All sorts of considerations become relevant as problems and so- Conclusion lutions. One important class of problems not easily The speculations at the end of the last section sug- considered under the science-as-explanation para- gest that there is a way of unifying the goals of phi- digm is finding new (mathematical losophy of science. The price is not negligible, methods, instruments, etc.) for producing known however. The deductive standard is given up in (including explanations themselves). On the favor of a new analytical tool with largely unknown old view, there would be no point in producing ad- . Moreover, the historical study of science ditional explanations of known results (except for can no longer be done independently of philosoph-

1 214 THE AMERICAN BIOLOGY TEACHER, VOLUME 47, NO. 4, APRIL 1985 ical analysis on this view, since the underlying dy- growthof knowledge.Cambridge: Cambridge namics of historical change are assumed to depend Press. Laudan, L. (1977). Progressand its problems.Berkeley: Uni- fundamentally on specific features of heuristic pro- versity of CaliforniaPress. cedures. With the reformulated goal of building hi- Lloyd, E. (1984). A semantic approach to the structure of erarchical (i.e., multiple level) models of scientific population . Philosophyof Science,51, 242-264. change, philosophy of science begins to look like a Nagel, E. (1961). The structureof science.New York: Har- science itself. court Brace. Nickles, T. (1973). Two of intertheoretic reduc- A scientific philosophy of science would be dis- tion. Journalof Philosophy,70, 181-201. comforting to those unprepared to look beyond their Popper, K. (1959). The logic of scientificdiscovery. : own for understanding, but it should liberate Hutchinson. others. Moreover, it brings home the point I made Popper, K. (1965). Conjecturesand refutations.2nd ed. New before about science -one can no longer York:. Shapere, D. (1964). The structure of scientific revolutions. be complacent with a fixed view of scientific method PhilosophicalReview, 73, 383-394. because the "science" which investigates "the" sci- Simon, H. (1957).Models of man. New York:. entific is on the move. Suppe, F. (Ed.). (1977). The structureof scientifictheories. 2nd ed. Urbana:University of Illinois Press. Suppes, P. (1967). What is a scientific theory? In S. Mor- References Downloaded from http://online.ucpress.edu/abt/article-pdf/47/4/211/86196/4448021.pdf by guest on 01 October 2021 genbesser (Ed.). Philosophyof sciencetoday. New York: Beatty, J. (1981). What's wrong with the received view of Basic Books. evolutionary theory? In P. Asquith and R. Giere (Eds.). Thompson, P. (1983). The structure of evolutionary theo- PSA 1980. Volume 2. East Lansing: Philosophy of Sci- ry: A semantic approach. Studiesin Historyand Philoso- ence Association. phyof Science,14, 215-229. Brown, H. (1977). Perception,theory and commitment.Ur- Toulmin, S. (1967). The evolutionary development of nat- bana: University of Illinois Press. ural science. AmericanScientist, 55, 456-471. Darwin, C. (1859). On the originof species.Cambridge: Har- Toulmin, S. (1971). From logical systems to conceptual vard University Press. (FacsimileReprint Edition). populations. In R. Buck and R. Cohen (Eds.). PSA 1970. Griesemer,J. (1984). Presentations and the status of theo- Dordrecht:D. Reidel. ries. In P. Asquith and P. Kitcher(Eds.). PSA 1984, Vol- Toulmin, S. (1972). Human understanding. Princeton: ume 1, East Lansing: Philosophy of Science Association. PrincetonUniversity Press. Hempel, C. (1965). Aspects of scientific explanation.New van Fraassen, B. (1980). Thescientific image. Oxford: Clar- York:Free Press. endon. Hempel, C. (1966). Philosophyof naturalscience. Englewood Williams, M. (1970). Deducing the consequences of evolu- Cliffs: Prentice-Hall. tion: A mathematicalmodel. Journalof TheoreticalBiology, Hull, D. (1975). Central subjects and historical narratives. 29, 343-385. Historyand theory,14, 253-274. Wimsatt, W. (1980). Reductionisticresearch strategies and Hull, D. (1982). The naked . In H. Plotkin (Ed.). their in the units of selection controversy. In T. Learning,development, and .London: John Wiley. Nickles (Ed.). Scientificdiscovery. Case studies. Dordrecht: Hull, D. (1983). Exemplarsand scientific change. In P. As- D. Reidel. quith and T. Nickles (Eds.). PSA 1982. Volume 2. East Wimsatt,W. (1981). Robustness, reliability,and overdeter- Lansing:Philosophy of Science Association. mination. In M. Brewer and B. Collins (Eds.). Scientific Kuhn, T. (1970). The structureof scientificrevolutions. 2nd inquiryand the socialsciences. San Francisco:Jossey-Bass. ed. Chicago:University of Chicago Press. Woodger, J. (1937). The axiomaticmethod in biology.Cam- Lakatos, I. & Musgrave, A. (Eds.). (1970). Criticismand the bridge: CambridgeUniversity Press.

A Salute to NABT's Sustaining Members American Optical Corpora- NABTthanks its sustaining The Pillsbury tion members for their Minneapolis,MN 55414. Buffalo,NY 14215.* continued support. Sargent-Welch Scientific Bausch and Lomb Sustainingmembers, Company Rochester,NY 14602. because of their special Skokie,IL 60076. Difco membership, receive Science Kit, Inc. Detroit, MI 48232.* discounts on advertising, Tonawanda,NY 14150. Lane Science Equipment exhibit and mailing Triarch Prepared Company list rentals. A sustaining Slides New York,NY 10122. membership is by Ripon, WI 54971.* Eli Lilly and Company invitation. Ward's Indianapolis,IN 46206. Establishment, Inc. NASCO, Inc. *indicatescharter sustaining Rochester,NY 14603. Fort Atkinson, WI 53538. membership