L. Berthet, C. DeDuve, Bull. soc. chim. biol.33, 21 (1951).

6. R. Wu and E. Racker, J. Biol. Chem. 234,1029 (1959).

7. F. Lynen, in Proceedings International Sym-posium on Enzyme Chemistry (Maruzen, To-kyo, 1958), p. 25.

8. B. Chance and B. Hess, Science 129, 700(1959); J. Biol. Chem. 234, 2421 (1959).

9. R. Balazs, Biochem. J. 72, 561 (1959).10. L. G. Abood, E. Bunngraber, M. Taylor, J.

Biol. Chem. 239, 1307 (1959).11. P. Hochstein, M. Woods, D. Burk, Federa-

tion Proc. 14, 227 (1955); P. Hochstein,Science 125, 496 (1957).

12. H. G. DuBuy and M. L. Hesselbach, J. Natl.Cancer Inst. 20, 403 (1958); J. Histochem. andCytochem. 4, 363 (1956).

13. A. C. Eisenberg, The Glycolysis and Respira-tion of Tumors (Academic Press, New York,1961).

14. N. 0. Kaplan, in The Enzymes, P. D. Boyer,H. Lardy, K. Myrback, Eds. (Academic Press,New York, 1960), vol. 3, pt. B, p. 105; C.Carruthers and V. Suntzeff, Cancer Research14, 29 (1954); S. Weinhouse, in Advances inCancer Research, J. P. Greenstein and A.Haddow, Eds. (Academic Press, New York,1955), vol. 3, p. 268.

15. A. L. Lehninger, J. Biol. Chem. 190, 345(1951).

16. B. Chance and G. R. Williams, Ibid. 217, 409(1955); G. F. Maley, ibid. 224, 1029 (1957).

17. C. Cooper and A. L. Lehninger, ibid. 219,489 (1956).

18. D. E. Green and F. L. Crane, in ProceedingsInternational Symposium on Enzyme Chemis-try (Maruzen, Tokyo, 1958), p. 275.

19. W. W. Kielley and J. R. Bronk, J. Biol. Chem.230, 521 (1958).

20. B. Chance and B. Sacktor, Arch. Biochem.Biophys. 76, 509 (1958).

21. T. M. Devlin and B. Bedell, in preparation.22. E. C. Slater and K. W. Cleland, Biochem. J.

53, 557 (1953).23. W. Chefurka, Enzymologfa 17, 73 (1954);

R. W. Estabrook and B. Sactor, J. Biol. Chem.233, 1014 (1958).

24. E. Zebe, A. Delbruck, T. Bucher, Biochem.Z. 331, 254 (1959).

25. G. E. Boxer and C. E. Shonk, Cancer Re-search 20, 85 (1960).

26. E. I. Ciaccio and D. L. Keller, FederationProc. 19, 34 (1960).

27. D. E. Green, D. M. Needham, J. G. Dewan,Biochem J. 31, 2327 (1937).

28. R. L. Ringler and T. P. Singer, Biochim. etBiophys. Acta 29, 661 (1958).

29. B. Sacktor, L. Packer, R. W. Estabrook, Arch.Biochem. Biophys. 80, 68 (1959).

30. M. Klingenberg and W. Slenczka, Biochem.Z. 331, 254 (1959).

31. E. I. Ciaccio, D. L. Keller, G. E. Boxer,Biochim. et Biophys. Acta 37, 191 (1960).

32. E. Leva and S. Rapoport, J. Biol. Chem. 149,47 (1943); T. Bucher and M. Klingenberg,Angew. Chem. 70, 552 (1958).

33. P. Holzer, P. Glogner, G. Sedlmayr, Biochem.Z. 330, 59 (1958).

34. A. Delbruck, H. Schimassek, K. Bartsch, T.Bucher, ibid. 331, 297 (1959).

35. B. Sacktor and A. Dick, Cancer Research 20,1408 (1960).

36. E. I. Ciaccio and J. B. Orange, Abstr. Am.Chem. Soc. Meeting, Div. Biol. Chem., Sept.1960, p. 27C.

37. H. P. Morris, H. Sidransky, B. P. Wagner,H. M. Dyer, Cancer Research 20, 1252 (1960).

38. M. Woods, personal communication.39. T. M. Devlin and B. Bedell, J. Biol. Chem.

235, 2134 (1960).40. , Federation Proc. 19, 34 (1960).41. A. L. Lehninger, H. C. Sudduth, J. B. Wise,

J. Biol. Chem. 235, 2450 (1960).42. H. A. Krebs, L. V. Eggleston, A. D'Alessan-

dro, Biochem. J. 79, 537 (1961).43. T. M. Devlin and B. Bedell, in preparation;

T. M. Devlin and G. E. Boxer, Proc. Am.Assoc. Cancer Research 3, 220 (1961).

44. A. Delbruck, E. Zebe, T. Bucher, Biochem. Z.331, 273 (1959).

45. T. E. Conover and L. Ernster, Biochem.Biophys. Research Communs. 2, 26 (1960).

46. N. 0. Kaplan, S. P. Colowick, E. F. Neufeld,J. Biol. Chem. 205, 1 (1953); N. 0. Kaplan,S. P. Colowick, L. J. Zatman, M. M. Ciotti,ibid. 205, 31 (1953).

47. A. M. Stein, N. 0. Kaplan, M. M. Ciotti,ibid. 234, 979 (1959).

48. T. M. Devlin, ibid. 234, 962 (1959).49. B. Reynafarje and V. R. Potter, Cancer Re-

search 17, 1112 (1957).50. P. Talalay and H. G. Williams-Ashman, Proc.

Natl. Acad. Sci. U.S. 44, 15 (1958).51. C. E. Wenner, M. A. Spirtes, S. Weinhouse,

Cancer Research 12, 44 (1952).52. H. Busch, ibid. 15, 365 (1955).53. - , M. H. Goldberg, D. C. Anderson,

ibid. 16, 175 (1956).54. H. Busch and P. V. Nair, J. Biol. Chem. 229,

377 (1957); J. Papaconstantinou and S. P.Colowick, J. Biol. Chem. 236, 278, 285(1961).

55. V. R. Potter, Federation Proc. 17, 691 (1958).

Cause and Effect in Biology

Kinds of causes, predictability, and teleologyare viewed by a practicing biologist.

Ernst Mayr

Being a practicing biologist I feel thatI cannot attempt the kind of analysis ofcause and effect in biological phenom-ena that a logician would undertake. Iwould instead like to concentrate on thespecial difficulties presented by the clas-sical concept of causality in biology.From the first attempts to achieve aunitary concept of cause, the student ofcausality has been bedeviled by thesedifficulties. Descartes's grossly mecha-nistic interpretation of life, and the log-ical extreme to which his ideas werecarried by Holbach and de la Mettrie,inevitably provoked a reaction leadingto vitalistic theories which have been invogue, off and on, to the present day.I have only to mention names likeDriesch (entelechy), Bergson (elan vi-

10 NOVEMBER 1961

tal), and Lecomte du Noily, among themore prominent authors of the recentpast. Though these authors may differin particulars, they all agree in claimingthat living beings and life processes can-not be causally explained in terms ofphysical and chemical phenomena. Itis our task to ask whether this assertionis justified, and if we answer this ques-tion with "no," to determine the sourceof the misunderstanding.

Causality, no matter how it is definedin terms of logic, is believed to containthree elements: (i) an explanation ofpast events ("a posteriori causality");(ii) prediction of future events; and(iii) interpretation of teleological-thatis, "goal-directed"-phenomena.

The three aspects of causality (ex-

planation, prediction, and teleology)must be the cardinal points in any dis-cussion of causality and were quiterightly singled out as such by Nagel (1).Biology can make a significant contri-bution to all three of them. But before Ican discuss this contribution in detail,I must say a few words about biologyas a science.

Biology

The word biology suggests a uniformand unified science. Yet recent develop-ments have made it increasingly clearthat biology is a most complex area-indeed, that the word biology is a labelfor two largely separate fields whichdiffer greatly in method, Fragestellung,and basic concepts. As soon as one goesbeyond the level of purely descriptivestructural biology, one finds two verydifferent areas, which may be designatedfunctional biology and evolutionary bi-ology. To be sure, the two fields havemany points of contact and overlap.Any biologist working in one of thesefields must have a knowledge and ap-preciation of the other field if he wants

The author is Alexander Agassiz professor ofzoology at Harvard University and director of theMuseum of Comparative Zoology, Harvard Col-lege, Cambridge, Mass. This article is adaptedfrom a lecture presented 1 February 1961 atMassachusetts Institute of Technology in the1960-61 series of Hayden lectures on "Causeand Effect."

1501

on

Janu

ary

2, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

to avoid the label of a narrow-mindedspecialist. Yet in his own research hewill be occupied with problems of eitherone or the other field. We cannot dis-cuss cause and effect in biology withoutfirst having characterized these twofields.

Functional biology. The functionalbiologist is vitally concerned with theoperation and interaction of structuralelements, from molecules up to organsand whole individuals. His ever-repeatedquestion is "How?" How does some-thing operate, how does it function?The functional anatomist who studiesan articulation shares this method andapproach with the molecular biologistwho studies *the function of a DNAmolecule in the transfer of genetic in-formation. The functional biologist at-tempts to isolate the particular compo-nent he studies, and in any given studyhe usually deals with a single individual,a single organ, a single cell, or a singlepart of a cell. He attempts to eliminate,or control, all variables, and he repeatshis experiments under constant or vary-ing conditions until he believes he hasclarified the function of the element hestudies. The chief technique of the func-tional biologist is the experiment, andhis approach is essentially the same asthat of the physicist and the chemist.Indeed, by isolating the studied phe-nomenon sufficiently from the complex-ities of the organism, he may achievethe ideal of a purely physical or chem-ical experiment. In spite of certainlimitations of this method, one mustagree with the functional biologist thatsuch a simplified approach is an ab-solute necessity for achieving his par-ticular objectives. The spectacular suc-cess of biochemical and biophysicalresearch justifies this direct, althoughdistinctly simplistic, approach.

Evolutionary biology. The evolution-ary biologist differs in his method andin the problems in which he is inter-ested. His basic question is "Why?"When we say "why" we must always beaware of the ambiguity of this term. Itmay mean "how come?," but it mayalso mean the finalistic "what for?" Itis obvious that the evolutionist has inmind the historical "how come?" whenhe asks "why?" Every organism, wheth-er individual or species, is the productof a long history, a history which in-deed dates back more than 2000 mil-lion years. As Max Delbriick (2) hassaid, "a mature physicist, acquaintinghimself for the first time with the prob-lems of biology, is puzzled by the cir-cumstance that there are no 'absolute

phenomena' in biology. Everything istime-bound and space-bound. The ani-mal or plant or micro-organism he isworking with is but a link in an evolu-tionary chain of changing forms, noneof which has any permanent validity."There is hardly any structure or func-tion in an organism that can be fullyunderstood unless it is studied againstthis historical background. To find thecauses for the existing characteristics,and particularly adaptations, of organ-isms is the main preoccupation of theevolutionary biologist. He is impressedby the enormous diversity of the or-ganic world. He wants to know the rea-sons for this diversity as well as thepathway by which it has been achieved.He studies the forces that bring aboutchanges in faunas and floras (as in partdocumented by paleontology), and hestudies the steps by which have evolvedthe miraculous adaptations so charac-teristic of every aspect of the organicworld.We can use the language of informa-

tion theory to attempt still another char-acterization of these two fields of biol-ogy. The functional biologist deals withall aspects of the decoding of the pro-grammed information contained in theDNA code of the fertilized zygote. Theevolutionary biologist, on the otherhand, is interested in the history ofthese codes of information and in thelaws that control the changes of thesecodes from generation to generation. Inother words, he is interested in thecauses of these changes.Many of the old arguments of biolog-

ical philosophy can be stated far moreprecisely in terms of these genetic codes.For instance, as Schmalhausen, in Rus-sia, and I have pointed out independ-ently, the inheritance of acquired char-acteristics becomes quite unthinkablewhen applied to the model of the trans-fer of genetic information from a pe-ripheral phenotype to the DNA of thegerm cells.

But let us not have an erroneousconcept of these codes. It is character-istic of these genetic codes that theprogramming is only in part rigid. Suchphenomena as learning, memory, non-genetic structural modification, and re-generation show how "open" these pro-grams are. Yet, even here there is greatspecificity, for instance with respect towhat can be "learned," at what stage inthe life cycle "learning" takes place, andhow long a memory engram is retained.The program, then, may be in partquite unspecific, and yet the range ofpossible variation is itself included in

1502

the specifications of the code. Thecodes, therefore, are in some respectshighly specific; in other respects theymerely specify "reaction norms" or gen-eral capacities and potentialities.

Let me illustrate this duality of codesby the difference between two kinds ofbirds with respect to "species recogni-tion." The young cowbird is raised byfoster parents-let us say, in the nestof a song sparrow or warbler. As soonas it becomes independent of its fosterparents it seeks the company of otheryoung cowbirds, even though it hasnever seen a cowbird before! In con-trast, after hatching from the egg, ayoung goose will accept as its parentthe first moving (and preferably alsocalling) object it can follow and become"imprinted" to. What is programmedis, in one case, a definite "gestalt," inthe other, merely the capacity to be-come imprinted to a "gestalt." Similardifferences in the specificity of the in-herited program are universal through-out the organic world.

Let us now get back to our maintopic and ask: Is cause the same thingin functional and evolutionary biology?Max Delbriick, again, has reminded

us (2) that as recently as 1870 Helm-holtz postulated "that the behavior ofliving cells should be accountable interms of motions of molecules actingunder certain fixed force laws." Now,says Delbriick correctly, we cannot evenaccount for the behavior of a singlehydrogen atom. As he also says, "anyliving cell carries with it the experiencesof a billion years of experimentation byits ancestors."

Let me illustrate the difficulties of theconcept of causality in biology by anexample. Let us ask: What is the causeof bird migration? Or more specifically:Why did the warbler on my summerplace in New Hampshire start his south-ward migration on the night of the 25thof August?

I can list four equally legitimatecauses for this migration.

1) An ecological cause. The warbler,being an insect eater, must migrate, be-cause it would starve to death if itshould try to winter in New Hampshire.

2) A genetic cause. The warbler hasacquired a genetic constitution in thecourse of the evolutionary history of itsspecies which induces it to respond ap-propriately to the proper stimuli fromthe environment. On the other hand,the screech owl, nesting right next to it,lacks this constitution and does not re-spond to these stimuli. As a result, it issedentary.

SCIENCE, VOL. 134

on

Janu

ary

2, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

3) An intrinsic physiological cause.The warbler flew south because its mi-gration is tied in with photoperiodicity.It responds to the decrease in day lengthand is ready to migrate as soon as thenumber of hours of daylight havedropped below a certain level.

4) An extrinsic physiological cause.Finally, the warbler migrated on the25th of August because a cold air mass,with northerly winds, passed over ourarea on that day. The sudden drop intemperature and the associated weatherconditions affected the bird, already ina general physiological readiness formigration, so that it actually took offon that particular day.Now, if we look over the four causa-

tions of the migration of this bird oncemore we can readily see that there is animmediate set of causes of the migra-tion, consisting of the physiological con-dition of the bird interacting withphotoperiodicity and drop in tempera-ture. We might call these the proximatecauses of migration. The other twocauses, the lack of food during winterand the genetic disposition of the bird,are the ultimate causes. These arecauses that have a history and that havebeen incorporated into the systemthrough many thousands of generationsof natural selection. It is evident thatthe functional biologist would be con-cerned with analysis of the proximatecauses, while the evolutionary biologistwould be concerned with analysis ofthe ultimate causes. This is the case withalmost any biological phenomenon wemight want to study. There is always aproximate set of causes and an ultimateset of causes; both have to be ex-plained and interpreted for.a completeunderstanding of the given phenomenon.

Still another way to express thesedifferences would be to say that proxi-mate causes govern the responses of theindividual (and his organs) to immediatefactors of the environment while ulti-mate causes are responsible for the evo-lution of the particular DNA code ofinformation with which every individualof every species is endowed. The logi-cian will, presumably, be little con-cerned with these distinctions. Yet, thebiologist knows that many heated argu-ments about the "cause" of a certainbiological phenomenon could have beenavoided if the two opponents had real-ized that one of them was concernedwith proximate and the other with ulti-mate causes. I might illustrate this by aquotation from Loeb (3): "The earlierwriters explained the growth of the legsin the tadpole of the frog or toad as a10 NOVEMBER 1961

case of adaptation to life on land. Weknow through Gudernatsch that thegrowth of the legs can be produced atany time even in the youngest tadpole,which is unable to live on land, by feed-ing the animal with the thyroid gland."

Let us now get back to the definitionof "cause" in formal philosophy andsee how it fits with the usual explana-tory "cause" of functional and evolu-tionary biology. We might, for instance,define cause as "a nonsufficient condi-tion without which an event would nothave happened," or as "a member of aset of jointly sufficient reasons withoutwhich the event would not happen"[after Scriven (4)]. Definitions such asthese describe causal relations quite ade-quately in certain branches of biology,particularly in those which deal withchemical and physical unit phenomena.In a strictly formal sense they are alsoapplicable to more complex phenom-ena, and yet they seem to have littleoperational value in those branches ofbiology that deal with complex systems.I doubt that there is a scientist whowould question the ultimate causalityof all biological phenomena-that is,that a causal explanation can be givenfor past biological eVents. Yet such anexplanation will often have to be so un-specific and so purely formal that itsexplanatory value can certainly be chal-lenged. In dealing with a complex sys-tem, an explanation can hardly be con-sidered very illuminating that states:"Phenomenon A is caused by a complexset of interacting factors, one of whichis b." Yet often this is about all onecan say. We will have to come back tothis difficulty in connection with theproblem of prediction. However, let usfirst consider the problem of teleology.

Teleology

No discussion of causality is completewhich does not come to grips with theproblem of teleology. This problem hadits beginning with Aristotle's classifica-tion of causes, one of the categoriesbeing the "final" causes. This categoryis based on the observation of the or-derly and purposive development of theindividual from the egg to the "final"stage of the adult, and of the develop-ment of the whole world from its be-ginnings (chaos?) to its present order.Final cause has been defined as "thecause responsible for the orderly reach-ing of a preconceived ultimate goal."All goal-seeking behavior has been clas-sified as "teleological," but so have

many other phenomena that are notnecessarily goal-seeking in nature.

Aristotelian scholars have rightly em-

phasized that Aristotle-by training andinterest-was first and foremost a bi-ologist, and that it was his preoccupa-tion with biological phenomena whichdominated his ideas on causes and in-duced him to postulate final causes inaddition to the material, formal, andefficient causes. Thinkers from Aristotleto the present have been challenged bythe apparent contradiction between amechanistic interpretation of naturalprocesses and the seemingly purposivesequence of events in organic growth, inreproduction, and in animal behavior.Such a rational thinker as Bernard (5)has stated the paradox in these words.There is, so to speak, a preestablished de-sign of each being and of each organ ofsuch a kind that each phenomenon byitself depends upon the general forces ofnature, but when taken in connection withthe others it seems directed by some in-visible guide on the road it follows andled to the place it occupies.We admit that the life phenomena are

attached to physicochemical manifesta-tions, but it is true that the essential is notexplained thereby; for no fortuitous com-ing together of physicochemical phenom-ena constructs each organism after a planand a fixed design (which are foreseen inadvance) and arouses the admirable sub-ordination and harmonious agreement ofthe acts of life. . . . Determinism cannever be [anything] but physicochemicaldeterminism. The vital force and life be-long to the metaphysical world.What is the x, this seemingly purpo-

sive agent, this "vital force," in organicphenomena? It is only in our lifetimethat explanations have been advancedwhich deal adequately with this paradox.The many dualistic, finalistic, and vi-

talistic philosophies of the past merelyreplaced the unknown x by a differentunknown, y or z, for calling an un-known factor entelechia or elan vitalis not an explanation. I shall not wastetime showing how wrong most of thesepast attempts were. Even though someof the underlying observations of theseconceptual schemes are quite correct,the supernaturalistic conclusions drawnfrom these observations are altogethermisleading.

Where, then, is it legitimate to speakof purpose and purposiveness in nature,and where is it not? To this questionwe can now give a firm and unambigu-ous answer. An individual who-to usethe language of the computer-hasbeen "programmed" can act purpose-fully. Historical processes, however, cannot act purposefully. A bird that startsits migration, an insect that selects its

1503

on

Janu

ary

2, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

host plant, an animal that avoids a

predator, a male that displays to a fe-male-they all act purposefully becausethey have been programmed to do so.

When I speak of the programmed "in-dividual," I do so in a broad sense. Aprogrammed computer itself is an "in-dividual" in this sense, but so is, duringreproduction, a pair of birds whose in-stinctive and learned actions and inter-actions obey, so to speak, a singleprogram.

The completely individualistic andyet also species-specific DNA code ofevery zygote (fertilized egg cell), whichcontrols the development of the centraland peripheral nervous systems, of thesense organs, of the hormones, of physi-ology and morphology, is the program

for the behavior computer of this in-dividual.

Natural selection does its best tofavor the production of codes guaran-

teeing behavior that increases fitness. Abehavior program that guarantees in-stantaneous correct reaction to a poten-tial food source, to a potential enemy,or to a potential mate will certainlygive greater fitness in the Darwiniansense than a program that lacks theseproperties. Again, a behavior program

that allows for appropriate learning andthe improvement of behavior reactionsby various types of feedbacks givesgreater likelihood of survival than a

program that lacks these properties.The purposive action of an individ-

ual, insofar as it is based on the prop-

erties of its genetic code, therefore isno more nor less purposive than theactions of a computer that has beenprogrammed to respond appropriatelyto various inputs. It is, if I may say so,

a purely mechanistic purposiveness.We biologists have long felt that it is

ambiguous to designate such pro-grammed, goal-directed behavior "tele-ological," because the word teleologicalhas also been used in a very differentsense, for the final stage in evolutionaryadaptive processes. When Aristotlespoke of final causes he was particularlyconcerned with the marvelous adapta-tions found throughout the plant andanimal kingdom. He was concernedwith what later authors have called de-sign or plan in nature. He ascribed tofinal causes not only mimicry or sym-

biosis but all the other adaptations ofanimals and plants to each other and totheir physical environment. The Aristo-telians and their successors asked them-selves what goal-directed process couldhave produced such a well-ordered de-sign in nature.

1504

It is now evident that the termsteleology and teleological have been ap-plied to two entirely different sets ofphenomena. On one hand is the pro-duction and perfecting throughout thehistory of the animal and plant king-doms of ever-new programs and of ever-improved DNA codes of information.On the other hand there is the testingof these programs and the decoding ofthese codes throughout the lifetime ofeach individual. There is a fundamentaldifference between, on the one hand,end-directed behavioral activities or de-velopmental processes of an individualor system, which are controlled by aprogram, and, on the other hand, thesteady improvement of genetic codes.This genetic improvement is evolution-ary adaptation controlled by naturalselection.

In order to avoid confusion betweenthe two entirely different types of enddirection, Pittendrigh (6) has intro-duced the term teleonomic as a descrip-tive term for all end-directed systems"not committed to Aristotelian teleol-ogy." Not only does this negative defini-tion place the entire burden on the wordsystem, but it makes no clear distinc-tion between the,, two teleologies ofAristotle. It would seem useful to re-strict the term teleonomic rigidly tosystems operating on the basis of a pro-gram, a code of information. Teleono-my in biology designates "the apparentpurposefulness of organisms and theircharacteristics," as Julian Huxley ex-pressed it (7).

Such a clear-cut separation of tele-onomy, which has an analyzable physi-cochemical basis, from teleology, whichdeals more broadly with the over-allharmony of the organic world, is mostuseful because these two entirely dif-ferent phenomena have so often beenconfused with each other.The development or behavior of an

individual is purposive, natural selectionis definitely not. When MacLeod (8)stated, "What is most challenging aboutDarwin, however, is his re-introductionof purpose into the natural world," hechose the wrong word. The word pur-pose is singularly inapplicable to evolu-tionary change, which is, after all, whatDarwin was considering. If an organismis well adapted, if it shows superiorfitness, this is not due to any purposeof its ancestors or of an outside agency,such as "Nature" or "God," who cre-ated a superior design or plan. Darwin"has swept out such finalistic teleologyby the front door," as Simpson (9) hasrightly said.

We can summarize this discussion bystating that there is no conflict betweencausality and teleonomy, but that scien-tific biology has not found any evidencethat would support teleology in the senseof various vitalistic or finalistic theories(9, 10). All the so-called teleologicalsystems which Nagel discusses (11) areactually illustrations of teleonomy.

The Problem of Prediction

The third great problem of causalityin biology is that of prediction. In theclassical theory of causality the touch-stone of the goodness of a causal ex-planation was its predictive value. Thisview is still maintained in Bunge's mod-ern classic (12): "A theory can predictto the extent to which it can describeand explain." It is evident that Bungeis a physicist; no biologist would havemade such a statement. The theory ofnatural selection can describe and ex-plain phenomena with considerable pre-cision, but it cannot make reliable pre-dictions, except through such trivial andmeaningless circular statements as, forinstance: "the fitter individuals will onthe average leave more offspring." Scri-ven (13) has emphasized quite correctlythat one of the most important contri-butions to philosophy made by theevolutionary theory is that it hasdemonstrated the independence of ex-planation and prediction.

Although prediction is not an In-separable concomitant of causality, ev-ery scientist is nevertheless happy if hiscausal explanations simultaneously havehigh predictive value. We, can dis-tinguish many categories of predictionin biological explanation. Indeed, it iseven doubtful how to define "predic-tion" in biology. A competent zoogeog-rapher can predict with high accuracywhat animals will be found on a previ-ously unexplored mountain range orisland. A paleontologist likewise canpredict with high probability what kindof fossils can be expected in a newlyaccessible geological horizon. Is suchcorrect guessing of the results of pastevents genuine prediction? A similardoubt pertains to taxonomic predictions,as discussed in the next paragraph. Theterm prediction is, however, surely legit-imately used for future events. Let megive you four examples to illustrate therange of predictability.

1) Prediction in classification. If Ihave identified a fruit fly as an individ-ual of Drosophila melanogaster on thebasis of bristle pattern and the propor-

SCIENCE, VOL. 134

on

Janu

ary

2, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

tions of face and eye, I can "predict"numerous structural and behavioralcharacteristics which I will find if Istudy other aspects of this individual. IfI find a new species with the diagnostickey characters of the genus Drosophila,I can at once "predict" a whole set ofbiological properties.

2) Prediction of most physicochem-ical phenomena on the molecular level.Predictions of very high accuracy canbe made with respect to most biochem-ical unit processes in organisms, such asmetabolic pathways, and with respectto biophysical phenomena in simplesystems, such as the action of light,heat, and electricity in physiology.

In examples 1 and 2 the predictivevalue of causal statements is usuallyvery high. Yet there are numerous othergeneralizations or causal statements inbiology that have low predictive values.The following examples are of this kind.

3) Prediction of the outcome ofcomplex ecological interactions. Thestatement, "An abandoned pasture insouthern New England will be replacedby a stand of grey birch (Betula populi-folia) and white pine (Pinus strobus)"is often correct. Even more often, how-ever, the replacement may be an almostsolid stand of P. strobus, or P. strobusmay be missing altogether and in itsstead will be cherry (Prunus), red cedar(Juniperus virginianus), maples, sumac,and several other species.Another example also illustrates this

unpredictability. When two species offlour beetles (Tribolium confusum andT. castaneum) are brought together ina uniform environment (sifted wheatflour), one of the two species will al-ways displace the other. At high tem-peratures and humidities, T. castaneumwill win out; at low temperatures andhumidities, T. confusum will be thevictor. Under intermediate conditionsthe outcome is indeterminate and henceunpredictable (Table 1) (14).

4) Prediction of evolutionary events.Probably nothing in biology is less pre-dictable than the future course of evo-lution. Looking at the Permian reptiles,who would have predicted that most ofthe more flourishing groups would be-come extinct (many rather rapidly), andthat one of the most undistinguishedbranches would give rise to the mam-mals? Which student of the Cambrianfauna would have predicted the revolu-tionary changes in the marine life ofthe subsequent geological eras? Unpre-dictability also characterizes small-scaleevolution. Breeders and students ofnatural selection have discovered again10 NOVEMBER 1961

Table 1. Two species of Tribolium in compe-tition [from Park (14)].

Condition Victorious speciesCondition_____ Replicas (N6. of trials)

Temp. Humidity (No.) T. con- T. cast-(OC) (%) fusum aneum

34 70 30 3029 70 66 1 1 5524 70 30 21 9

34,29 30 60 53 724 30 20 20

and again that independent parallellines exposed to the same selection pres-sure will respond at different rates andwith different correlated effects, noneof them predictable.As is true in many other branches of

science, the validity of predictions forbiological phenomena (except for afew chemical or physical unit processes)is nearly always statistical. We can pre-dict with high accuracy that slightlymore than 500 of the next 1000 new-borns will be boys. We cannot predictthe sex of a particular unborn child.

Reasons for Indeterminacy in Biology

Without claiming to exhaust all thepossible reasons for Indeterminacy, Ican list four classes. Although theysomewhat overlap each other, each de-serves to be treated separately.

1) Randomness of an event with re-spect to the significance of the event.Spontaneous mutation, caused by an"error" in DNA replication, illustratesthis cause for indeterminacy very well.The occurrence of a given mutation isin no way related to the evolutionaryneeds of the particular organism or ofthe population to which it belongs. Theprecise results of a given selection pres-sure are unpredictable because muta-tion, recombination, and developmentalhomeostasis are making indeterminatecontributions to the response to thispressure. All the steps in the determi-nation of the genetic contents of azygote contain a large componentof this type of randomness. What wehave described for mutation is alsotrue for crossing over, chromosomalsegregation, gametic selection, mateselection, and early survival of thezygotes. Neither underlying molecularphenomena nor the mechanical mo-tions responsible for this randomnessare related to their biological effects.

2) Uniqueness of all entities at thehigher levels of biological integration.In the uniqueness of biological entitiesand phenomena lies one of the major

differences between biology and thephysical sciences. Physicists and chem-ists often have genuine difficulty in un-

derstanding the biologist's stress of theunique, although such an understandinghas been greatly facilitated by the de-velopments in modern physics. If a

physicist says "ice floats on water," hisstatement is true for any piece of iceand any body of water. The membersof a class usually lack the individualitythat is so characteristic of the organicworld, where all individuals are unique;all stages in the life cycle are unique;all populations are unique; all speciesand higher categories are unique; allinterindividual contacts are unique; allnatural associations of species are

unique; and all evolutionary events areunique. Where these statements are ap-plicable to man, their validity is self-evident. However, they are equally validfor all sexually reproducing animalsand plants. Uniqueness, of course, doesnot entirely preclude prediction. Wecan make many valid statements aboutthe attributes and behavior of man, andthe same is true for other organisms.But most of these statements (exceptfor those pertaining to taxonomy) havepurely statistical validity. Uniqueness isparticularly characteristic for evolu-tionary biology. It is quite impossibleto have for unique phenomena generallaws like those that exist in classicalmechanics.

3) Extreme complexity. The physi-cist Elsaisser stated in a recent sympo-sium: "[an] outstanding feature of allorganisms is their well-nigh unlimitedstructural and dynamical complexity."This is true. Every organic system is sorich in feedbacks, homeostatic devices,and potential multiple pathways that acomplete description is quite impossible.Furthermore, the analysis of such a sys-tem would require its destruction andwould thus be futile.

4) Emergence of new qualities athigher levels of integration. It wouldlead too far to discuss in this contextthe thorny problem of "emergence."All I can do here is to state its principledogmatically: "When two entities arecombined at a higher level of integra-tion, not all the properties of the newentity are necessarily a logical or pre-dictable consequence of the propertiesof the components." This difficulty isby no means confined to biology, butit is certainly one of the major sourcesof indeterminacy in biology. Let us re-member that indeterminacy does notmean lack of cause, but merely un-predictability.

1505

on

Janu

ary

2, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

All four causes of indeterminacy, in-dividually and combined, reduce theprecision of prediction.One may raise the question at this

point whether predictability in classicalmechanics and unpredictability in biol-ogy are due to a difference of degree orof kind. There is much to suggest thatthe difference is, in considerable part,merely a matter of degree. Classicalmechanics is, so to speak, at one endof a continuous spectrum, and biologyis at the other. Let us take the classicalexample of the gas laws. Essentiallythey are only statistically true, but thepopulation of molecules in a gas obey-ing the gas laws is so enormous that theactions of individual molecules becomeintegrated into a predictable-one mightsay "absolute"-result. Samples of fiveor 20 molecules would show definiteindividuality. The difference in the sizeof the studied "populations" certainlycontributes to the difference betweenthe physical sciences and biology.

Conclusions

Let us now return to our initial ques-tion and try to summarize some of ourconclusions on the nature of the cause-and-effect relations in biology.

1) Causality in biology is a far cryfrom causality in classical mechanics.

2) Explanations of all but the sim-plest biological phenomena usually con-sist of sets of causes. This is particularlytrue for those biological phenomenathat can be understood only if theirevolutionary history is also considered.Each set is like a pair of brackets whichcontains much that is unanalyzed andmuch that can presumably never beanalyzed completely.

3) In view of the high number ofmultiple pathways possible for mostbiological processes (except for thepurely physicochemical ones) and inview of the randomness of many of thebiological processes, particularly on themolecular level (as well as for otherreasons), causality in biological sys-tems is not predictive, or at best is onlystatistically predictive.

4) The existence of complex codes ofinformation in the DNA of the germplasm permits teleonomic purposive-ness. On the other hand, evolutionaryresearch has found no evidence what-soever for a "goal-seeking" of evolu-tionary lines, as postulated in that kindof teleology which sees "plan and de-sign" in nature. Thae harmony of theliving universe, so far as it exists, is ana posteriori product of natural selection.

Finally, causality in biology is not inreal conflict with the causality of clas-sical mechanics. As modern physics hasalso demonstrated, the causality of clas-sical mechanics is only a very simple,special case of causality. Predictability,for instance, is not a necessary com-ponent of causality. The complexities ofbiological causality do not justify em-bracing nonscientific ideologies, such asvitalism or finalism, but should en-courage all those who have been tryingto give a broader basis to the conceptof causality.

References and Notes

1. E. Nagel, lecture presented at the Massachu-setts Institute of Technology in the 1960-61Hayden Lectures series.

2. M. Delbrilck, Trans. Conn. Acad. Arts Scl.38, 173 (1949).

3. J. Loeb, The Organism as a Whole (Putnam,New York, 1916).

4. M. Scriven, unpublished manuscript.5. C. Bernard, Lecons sur les phinomMnes de la

vie (1885), vol. 1.6. C. S. Pittendrigh, in Behavior and Evolution,

A. Roe and G. G. Simpson, Eds. (Yale Univ.Press, New Haven, Conn., 1958), p. 394.

7. J. Huxley, Zool. Jahrb. Abt. Anat. u. Ontog.Tiere 88, 9 (1960).

8. R. B. MacLeod, Science 125, 477 (1957).9. G. G. Simpson, Ibid. 131, 966 (1960).

10. , Sc. Monthly 71, 262 (1950); L. F.Koch, Ibid. 85, 245 (1957).

11. E. Nagel, The Structure of Science (HarcourtBrace and World, Inc., New York, 1961).

12. M. Bunge, Causality (Harvard Univ. Press,Cambridge, Mass. 1959), p. 307.

13. M. Scriven, Science 130, 477 (1959).14. T. Park, Physlol. Zool. 27, 177 (1954).

Development of molecular amplifiersin the visible and near-visible region(1) of the electromagnetic spectrumhas been in progress at several labora-tories. Such amplifiers go under thedesignation of "laser" or optical maser,the former term being an acronym forThe authors are affiliated with TRG, Inc.

(Technical Research Group), Syosset, N.Y.

1506

light amplification by stimulated emis-sion of radiation. Such devices havebeen successfully demonstrated at sev-eral places (2), and several industrialorganizations have made them availablecommercially. It is almost certain thatlasers will be incorporated into com-munications and other technologies ata rapid rate. This article presents some

preliminary calculations which are ofphysiological interest in terms of thehazard associated with laser beams andtheir potential employment as biologicaland clinical tools.

Properties

From the point of view of physio-logical interest there are two importantproperties of laser beams, the extremelycollimated character of the light and itshigh degree of monochromaticity. Thecollimation property implies the possi-bility of obtaining large energy densitiesin narrow beams. The optimum diver-gence angle of a laser beam, O,min, islimited only by the wavelength of thelight emitted and the diameter of thelaser source in accord with the Fraun-hofer diffraction relationship:

Om.n= 2.44X/DL (1)

Here X is the wavelength of the emittedlaser light and DL is the diameter ofthe beam emerging from the laser sourceor from a subsequent lens system, ifone is used.

SCIENCE, VOL. 134

Physiological Implicationsof Laser Beams

The very high radiation flux densities of opticalmasers point to important biomedical applications.

Leonard R. Solon, Raphael Aronson, Gordon Gould o

n Ja

nuar

y 2,

201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from