systems theory philosophy phillips 1969

8/12/2019 Systems Theory Philosophy Phillips 1969

1/14

D . C. PHILLIPS

Systems Theory Discredited PhilosophyModem systems theory appears to have developed as a result of dissatisfactionfelt in certain quarters with the traditional method of studying complex systems.This traditional analytic or mechanistic method was to div ide. a system intodiscrete parts each of which was studied in isolation from the others. The wholesystem was then regarded as being the sum total of these isolated parts; accordingto this mechanistic view, it was the parts which determined the nature of the wholesystem. As an illustration a supporter of the mechanistic method might well choosethe example of a watch. The features of the parts of the watch determine, so theargument runs, the features of the watch as a wholefor example, the accuracyand weight of the whole watch are determined by the precision and weight ofthe parts.

One of the founders of what is now known as general system theory wasLudwig von Bertalanffy. It is not surprising that he started his career as a biologist;for it is in biology, particularly, that an apparently good case can be made g instthe mechanistic or analytic method, and for the opposing organismic or systemsv/ew.i Consider, as an example of a complex biological system, the kangaroc. Itcould be argued by an organicist or system theorist that the features of the partsof this organism or system are determined by the characteristics of the wholeorganismfor example, the characteristics of the digestive tract of the kangarooare determined by the nature of the whole kangaroo, his habitat, dietary habits,physiological features, heredity, evolutionary history, and so on.But the organicist would not stop here. He would proceed to attack themechanist s example by claiming that the treatment of a watch as an aggregate ofparts is inadequate because it omits two essential features of the watch, namely,

the order or interrelation of the parts, and the purpose that the watch was designedto fulfil. Any account of a watch that does not include these aspects is incomplete.Finally, apparently clinching his case, the organicist would argue that the theorywhich fails to describe a watch adequately, is utterly hopeless when faced withthe task of describing a complex living system. This point is well illustrated by apassage written by Bertalanflfy in 1933:1. These points are developed in m ore detail in D. C . Phillips, Organicism in the L ateNineteen th and Early T wentieth C enturies , which is to be pu blished in The Journat of theHistory of IdeasD . C. PHILLIPS is a Lecturer in Education in Monash University. This paper waspresented at the Fifth Canberra Seminar on Administrative Studies, Australian NationalUniversity, May 1969.


2/14

ABACUSMechanism . . . provides us with no grasp of the specific characterist ics of organisms,of the organization of organic processes among one another, of organic 'wholeness' ,of the problem of the origin of organic ' teleology' , or of the historical character oforganism s. . . . W e must therefore t ry to establ ish a new standpoint which asopposed to mechanism takes account of organic wholeness, but . . . t reats i t ina manner which admits of scientific investigation.^The new 'standp oint' which Bertalanfiy called for in order to do justice to 'organicwholeness' was of course general system theory (GST). In a more recent work.Problem s of Life, Bertalanffy called for the full and rigorous development of theorganismic view, for he believed that organismic principles were applicable notonly to the study of biological systems, but to allsystems. Bertalanffy wrote :From the statements we have made, a stupendous perspective emerges, a vista towardsa hitherto unsuspected unity of the conception of the world. Similar general principleshave evolved everywhere, whether we are dealing with inanimate things, organisms,mental or social processes. What is the origin of these correspondences?We answer this question by the claim for a new realm of science, which w e callGeneral System Theory. It is a logico-mathematical field the subject matter ofwhich is the formulation and derivation of those principles which h old for systems ingeneral. A system can be defined as a complex of elements standing in interaction.There are general principles holding for systems, irrespective of the nature of thecomponent elements and of the relations or forces between them.^This call for the development of general system theory evidently was successful,for a considerable body of literature has developed in the field; an even greatermass of material has accumulated around the borders of thefield forexample,

William G. Scott has written that 'Modem organization theory is on the peripheryof general system theory.'The foregoing brief discussion will serve as an introduction to general systemtheory. In the following pages five features of GST have been selected forexamination, namely:1. the failure of systems theorists to appreciate the history of their theory;2. the failure to specify precisely what is meant by a 'system ';3. the vagueness over what is to be included w ithin systems theory;4. the weakness of the charges brought against the analytic or mechanisticmethod;5. the failure of GST as a scientific theory .

1. The failure of systems theorists to ppreci te the history of their theoryIt is easy to find references, in recent writings on systems theory, to the workof Bertalanffy and to the shortcomings of the analytic method when applied tocomplex systems such as biological organisms. But it is rare to come across anymore detailed account of the history of the central theses of GST.2. Ludwig von Bertalanffy, Modern Theories of Development, transl. by J. H. Woodger,Harper Torchbooks, New York 1962, p. 46.3. Ludwig von Bertalanffy, Problems of Life, Harper Torchbooks, New York 1960, p. 199.(My emphasis)4. William G. Scott, 'Organization Theory: An Overview and Appraisal', in Joseph A. Litterer


3/14

SYSTEMS THEORY A DIS REDITED PHILOSOPHYOne important insight into this history was revealed by Stafford Beer in a paperpresented at the Firs t Systems Symposium at Case Institute ofTechnology in1960.5 Beer commented that some of the ideas of the German philosopher Hegel(1770-1831) were relevant to systems theory, but he did not develop the point

in any detail. The point is worth pursuing^ however, for Hegelian ideas certainlyare relevant to systems theory. Hegel heldaform of organicism; he was an earlybut great systems theorist. But the interesting point is that his philosophy has longbeen discredited. The weaknessesofHegelian thought are tosome extent theweaknesses of GST.Hegel is the most difficult, the most obscure, ofall the major philosophers toread, but despite this (or because ofit) he has been extremely influential. Hisphilosophy dominated the English-speaking world of the late nineteenth and earlytwentieth centuries, and it isthe H egelian philosophy current at this time tbatwill be referred to in what follows.The Hegelians regarded the wholeof reality as forming asystem the parts ofwhich were organicallyorinternally interrelated. Being a system, reality couldnot be studied successfully by dividingit into pa rts each of which was studied inisolation. Forwhen a pa rt was isolated from thewhole system its naturechangedit was no longer a part of the whole, and it became an inaccurate guideas to the natureofthe whole.It is apparent, therefore, that the Hegelian theoryof organic or internal relations was directly opposed to the analytic or mechanisticmethod.*The Hegelian theory of relations does not stand up to investigation. The theoryis based upon the supposition that entities (such as the parts of asystem) arealtered by the relationships into which they enter.If A, B, and Care the inter-related partsof a system, then the natures ofA, B, and Cmust be affected bythe interrelationships. Without this supposition, there would be no objection toisolating the parts of a system and studying them separately. The Oxford Hegelianphilosopher, F. H. Bradley, writing inthe late nineteenth century, gave a fairlyclear account of the theory of internal relations. He maintained that when entityA entered into a relationship with entity B or C it gained some property or qualityor characteristicPas aresult ofthis relationship. W ithout the relationship, and

hence without the property P, Bradley argued, A would be different, itwould benot-A. Any relation at all between A and any other entity necessarily determinedsome property of A, without which A would be different from what it is. This wasthe heart of the theory of internal relations: entities necessarily are altered by therelations into which they enter. Or as Bradley summed u p the theory, And therelation also must penetrate the inner being of its terms. * To mechanists, on the5. Stafford Beer, Below the Twilight ArchA Mythology ofSystems inDonald P. Eckman(ed .) . Systems: Research and Design John Wiley, New York 1961.6. For further discussion see Phillips, Organicism . . . .7. See for example, F. H. Bradley, Appearance and Reality Oxford University Press, 1962.pp. 513-19.


4/14

CUSOther hand, relations do not alter the entities that are related; as Hegel himselfobserved:

And this constitutes the characteristic of Mechanism, which is, that whatever relationrelates the terms is foreign' to them and does not concern their nature; even if itinvolves the appearance of a One, it remains nothing else than a collocation, mixture,heap, or the like.'The Hegelians' case has force so long as discussion centres around the 'nature' of

an entity, but much of the force is lost if the theory of internal relations isexpressed in the terminology of the mid-twentieth century. The emphasis on thenature of entities is replaced by emphasis on the defining and accom panyingcharacteristics affecting the usage of terms. Every entity has an indefinitely largenumber of characteristics, and those characteristics without which the entity wouldnot be designated by a certain term are the characteristics that define the term.Accompanying characteristics are those characteristics that are not definingtheirpresence or absence makes no difference to the use of the term.'

Interpreted in this light, tbe Hegelians appear to have been maintaining that asa result of its relationship with any other entity B or C, entity A would havesome characteristic P, arid furthermore this characteristic would be one of thedefining characteristics of A. Without A's relationship to entity B or C, thisdefining characteristic P would not exist, and thus in fact A would be not-A.Every relationship into which A entered, no matter what sort of a relationship itwas, would determine a defining characteristic of A.

When phrased in terms of defining characteristics, the theory of internalrelations can be seen to face serious difficulties. The main difficulty is that not allthe characteristics of an entity are defining characteristics; many of the character-istics are accompanying characteristics. Thus, even if it is admitted that everyrelation which A enters into determines some characteristic of A, it is notnecessarily the case that the characteristic determined by any one of these relationswill be a defining characteristic of A. The characteristics determined by some ofthe relations wiU be defining characteristics, bu t some may no t b e . It is possible,therefore, for A to enter into a relationship and yet remain unchanged.

A second difficulty in the theory of internal relations is that it makes the attain-ing of knowledge impossible. For to have knowledge of A, in the sense of knowingthe defining features of A, all of A's relationships would have to be known; butsince A is related to everything else in the whole of which it is a part, this wholemust be known before A can be known. William James, who was a contemporaryof Bradley, brilliantly parodied this particular feature of Hegelian thought:9. G. W. F. Hegel, TheScience of Logic transl. by Johnston and Struthers, Vol. II, Allen &Unwin, London 1929, p. 350. See alsoG. E. Moore, 'External and Internal Relatidns' inhis hilosophical Studies Routledge and Kegan Paul, London 1960, especially p. 284.10. John Hospers, An Introduction to Philosophical Analysis rev. edn,Routledge and KeganPaul, London 1967, Ch . 1.


5/14

SYSTEMS THEORY A DtSCREDlTED PHILOSOPHYIt costs nothing, not evenam ental eflfort, to ad m it that the abs olute totalityofthingsm ay beorgan ized exac tly after the pa tternof one of these through-and-throughabstractions. Infact, it isthe p leasantest and freestofmental movem ents. H usbandmakes, and is made by, wife, through marriage; one makes other by being itself other;everything self-created through itsopposite you go round like a squirrel in acage.. . . Wha t in fact is the logicofthese abstract systems? It is as we said above:if any Mem ber then theWh ole System; if not theWho le System then h^^

Or as Bertrand Russell saidofHegelianism more recently, If all knowledge wereknowledgeofthe universe asa whole, there would be no knowledge.''^The theory of internal relations had at least four corollaries that were recognizedby Hegelian philosophersofthe late nineteenth century (and, indeed, probablyby everyone who accepted organismic ideas). They were:(i) the whole is more than the sum ofthe parts; (ii) the whole determines thenature of the parts; (iii) the parts cannot be understood ifconsidered in isolationfrom the whole; (iv) the parts are dynamically interrelated or interdependent.These ideas can be criticized on anumber of grounds.'^ In the first place, beingbased upon thetheory of internal relations, they become suspect as soonasthis theory is rejected. T he fourth idea, however, can besupported onothergrounds, and removal of the support givenby the theory of internal relationsis not fatal to ita mechanist, forexample, can fully agree that the partsofcertain systems are dynamically interrelated. Secondly, the first three ideassuffer from vagueness,or worse. The statement of the first idea, forinstance,uses the term 'sum', without making clear what precisely ismeant. In mathe-

matics there are three types of 'sum'arithmetic sum, algebraic sum,andvector sum;it isnota difficult matter toinvent and stipulate a senseof 'sum'in which thewhole s greater than the sum of theparts. Thesecond ideasuggests that, because the parts are parts of thewhole, thewhole determinesitself.Asone critic wrote in 1903: 'That this supposition is self-contradictorya very little reflection should besufficient to show .' The third idea suggeststhat it cannot bepredicted, from thefact that anentity hasproperties 1, m,andnwhenit isisolated from certain other entities, what properties the entitywill have whenit comes into relationships with these other entities. This sugges-tion isunw arranted; the Hegelians and organicists have offered noevidencetoshow that this typeofprediction isimpossible. Furtherm ore, the weightof theevidence actually is against them here, for thephysical sciences canprovidenumerous examples ofsuccessful prediction of the properties anentity canbeexpected to display whenit isplaced innovel conditions.'^12. William James, 'Absolutism and Empiricism' in Mind IX, 1884, pp. 282-3. (My emphasis)13. Bertrand Russell, HistoryofWestern Philosophy Allen Unwin, London 1948, p. 772.14.A detailed discussionof the four ideas, and the criticisms that canbelevelled againstthem,isgiven in Phillips, 'Organicism. . . .15. G. E. Moore, Principia Ethica London 1960, p. 33.16. For example,itis possible to predict the properties materials will possess under abnormalconditionsof temperatureorpressure,orthe characteristicsa component will display when


6/14

ABACUSApart from Hegelian philosophy, there is one further aspect of the history ofthe ideas underlying general system theory that recent writers have ignored.This concems developments within biological science in the late nineteenthcentury. Many of the points made by Bertalanffy in the decades since 1930

were given a prior statement in this earlier period. In the late nineteenthcentury, findings in embryology, cellular biology, and physiology, led a numberof biologists to maintain that the mechanistic method was, in principle,inapplicable to biology. The diflRculty was tha t the biologists were unab le todevelop mechanistic theories to explain the new findings. (It can now be seenthat their knowledge of biology, and of relevant related disciplines such asbiochemistry, was not detailed enough to enable them to produce successfulexplanations.) Thus, writing in 1884, J. S. Haldane rejected the mechanisticapproach; he produced this remarkably contemporary-sounding systems-typestatement:These parts (of the organism) stand to one another and to the surroundings, not inthe relation of cause and effect, but in that of reciprocity. The parts of an organismand its surroundings thus form a system, any one of the parts of which constantlyacts on the rest, but only does so, qu a part of the system, in so far as they at thesame time act on it.'*

J. S. Haldane, together with his brother R. B., and Edmund Montgomery,were prominent organicists in the late nineteenth century; they were joined bymany others in the twentieth century, including J. H. Woodger, C. LloydMorgan, E. R. Russell, W. E. Agar, and the statesman. General Smuts.In conclusion, it must be re-emphasized that the ideas that have been discussedin this account of the Hegelianism and organismic biology of the late nineteenthand early twentieth centuries are ideas that are often referred to by contem-porary writers on systems theory. One could be excused for mistaking thefollowing mid-twentieth-century argument from Bertalanify, for example, for asample of late nineteenth-century thought:Every organism represents a system, by which term we mean a complex of elementsin mutual interaction. From this obvious statement the limitations of the analyticaland summative conceptions must follow. First, it is impossible to resolve thephenomena of life completely into elementary units; for each individual part andeach individual event depends not only on conditions within itself but also to agreater or lesser extent on the conditions within the whole, or within superordinateunits of which it is a part. Hence the behaviour of an isolated part is, in general,different from its behaviour within the context of the whole. . . . Secondly, the actualwhole shows properties that are absent from its isolated parts.

2 . The failure to specify precisely what is meant by a systemGST has as its function the formulation of the laws applicable to all systems,

irrespective of what actually composes the elements of these systems. Although17. This led to a resurgence of vitalism in biology in the late nineteenth century, togetherwith the development of organicism and the 'theory' of creative evolution.18. J. S. Haldane, 'Life and Mechanism' inMind IX, 1884, p. 33.


7/14

SYSTEMS THEORY A DISCREDITED PHILOSOPHYsomewhat ambitious, the quest forGST isnot necessarily Qu ixotic; alldependsupon w hat is meant by the term system . Butitis here th at the accoun ts given b ysystems theorists have been unsatisfactory.

Bertalanffy defined a system as a complex ofelements inmutual interaction .Anatol Rapoport s tated thata whole which functions as awho le by virtue oftheinterdependence of its parts iscalled a system . . . .^^ And R. L.Ackoff wro tetha t initially we can define a system broad ly and crudely as any entity, con-ceptual orphy sical, w hich consists of interdep ende nt par ts .^ Th ese definitionsshow that no significant advance in thinking about the general conceptof asystemhas taken place since the late nineteenth century; the Haldane brothers and thebiologist Edmund Montgomery hadideas that ranparallel to those of mid-twentieth century systems theorists. It is instructive to examine the situationthe men of thelate nineteenth century found themselves in because of thisnotion of a system.

Edmund Montgomery hadmade a particular study of s impler animals; hereached the conclusion that in allanimals the whole (orthe system) is herein all reality anteceden t to itspar ts . Th e par ts of anorganism or system arespecialized and segregated from a pre-existing w hole, an d are inno way discreteand independent units joined together . . . . ^Th e Haldane brothers wentfurther. J. S. Hald ane wrote that the w hole was not merely the organism , butthe organism together with theenvironm ent; then with his brother he tookanother step forward; they suggested that it followed from thefact that anorganism and its environment form a system, that thespec ies m ay itself belooked upon as a com poun d organism .^ Th e H egelian p hilosophers took thefinal s tep; all of reality (or theAb solute, as they called it) was a whole, asystem. In F. H.Bradley s wo rds :The universeisone inthis sense that its differences exist harmoniously within onewhole, beyond which thereisnothing. H ence the Absolute is, so far, an individualandasystem.. . .^It is app aren t that the same reasoning underlies these various exam ples; theprinciples that ledMontgomery to identify thewhole as being the individualbiological organism, led theHaldanes and the Hegelians to their respectiveviews. An exactly similar situatio n exists for general system theorists. Theconcept of a system with which they opera te does not give them any grou ndsfor limiting their attention toanything bu t the universe as a whole. An example

20. Anatol Rapoport , Foreword in W alter Buckley (ed .), Mo dern System.^ Researchforthe Behavioral Scientist Chicago 1968, p. xvii.21.R. L.Ackoff Systems, Organizations, and Interdisciplinary Research inEckman (ed. ) .Systems . . . pp. 27-8.22 . Edmund Montgomery, The Unity of the O rganic Individual inMind V,1880, p.326.23.R. B. and J. S. Haldane, The R elation of Philosophy toScience inAndrew Seth andR. B. Haldane (eds). Essays in Philosophical Criticism London 1883, p. 58.


8/14

CUS

should make the point clear. Suppose a general system theorist is treating anarmy unit as a system. Now, if a system is a 'complex of elements in interaction',it can be argued with some justice that an army unit is not an independentsystem as it is itself interrelated with many other 'entities' to form a moreembracing system. And once the general system theorist embarks on this road,as in fact he did when he defined a system as 'elements in interaction', heseems committed to continuing his journey until he arrives at the end of theroad and joins the Hegelians in contemplating the whole of reality Th isdifficulty has been recognized by at least one systems theorist; Stafford Beer,in his paper to the Symposium at Case Institute of Technology, stated that:

. . . earlier it was contended that the boundaries of a system are subjective; andthis is strongly supported at the philosophical level by the Hegelian axiom ofinternal relationswhich of course makes it logically possible to equate everysystem with the universe itself So the crucial scientific problem for systemsresearch is this: how to separate a particular viable system for study from the restof the universe without committing an annihilating divisio these are problemsof desperate urgency for every nontrivial systems study. . . .^^Beer was not exaggerating when he identified this problem as being of desperateurgency; and he also was right in raising the issue of triviality. It is indeed trivialto define a system in terms of interrelation of components; every entity in theuniverse enters into some relationships, and everything, therefore, can be regardedas a com pon ent of some sort of system.^* T he genera l system the orist s' definitionof a system, seen in this light, is not particularly informative. In effect the definitiontells us that a system is that which can be composed of any things interrelatedanyhow; for sheer profundity this bears comparison with William James' famousparody of the definition of evolution:Evolution is a change from a no-howish untalkaboutable all-alikeness to a some-howish and in general talkaboutable not-all-alikeness by continuous sticktogetherationsand somethingelseifications.^''It might be thought that the general system theorists' problem of specifyingwhat they mean by a system is susceptible of a simple solution. Camiot a theorist

simply select out certain interrelated entities that happen to be of relevance tohis particular investigation, and call this group of entities a system? This is pro-posed, for example, by A. D. Hall and R. E. Fagen, in the introductory chapter25. Stafford B eer, Below the Tw ilight A rch , pp. 18-19. It is interesting to note that in awork on systems engineering Hall and Fagen raise the same problem in another form. Theydefine the environment of a system as that which is in interrelation with the system, andthis leads them to the problem of distinguishing when an object belongs to a system andwhen it belongs to the environment. They state: The answer is by no means definite.A. D. Hall and R. E. Fagen, Definition of System in Buckley ( ed .) . Modern SystemsResearch p. 83.26. H all and Fagen acknow ledge this by stating For any given set of objects it is impossibleto say that no interrelationships exist . . . , Definition of System , p. 82.27. As quoted by Ralph Barton Perry, The Thought and Character of Wtltiam James Vol. I,


9/14

SYSTEMS THEORY A DIS REDITED PHILOSOPHYof SystemsEngineering.^ But this cure for the systems theorists' ills turns out tobe far worse than the original disease, for it runs foul of some of the fundamentalprinciples of GST. In the first place, the theorist can only select relevant entitiesif he has some principlesofselectionorcriteriaofrelevance, and he can onlyhave these for the particular problem under considerationif investigation has beenunder way and has made considerable headway. The investigation must havereached the stage where it is possible to say entities a, b, and care relevant to thesolution of the problem and entities d, e, andfare not relevant. Furthermore, thisfairly advanced stageofinvestigation must have been reached without using sys-tems methods, for what is under consideration here is what is necessary beforeasystems theorist can select or define his system. In fact,itis most likely that themechanistic or analytic method that general system theorists claim is not applicable,has been the method used to achieve these important advances in the investigation.The second objection to the simple cure that was proposed, is that in stipulatingthat the particular group of entities of interest to him form the relevant system, thegeneral system theorist has not gonefarenough. H e also hasto show thatinisolating this system from the other entities with whichitis normally interrelated,he has not introduced artifacts; or as Beer put it, the systems theorist has to showthat he has not committed an 'annihilating divisio\ One of the main points madeby GSTisthat the interrelationships between the partsof a system areofvitalimportance, and inisolating a system forstudy the theorist issevering someinterrelationshipshe is doing the very thing that his own creed tells him shouldnot be done.3. The v gueness over what is to be included within systems theory

In Bertalanffy's early conception, GST had as its subject matter 'the formulationand derivationofthose principles which hold for systems in general'.^' In subse-quent years, however,apuzzling situation developed. Some works on GST haveput forward cl ssific tions of systems,' while others have contained discussions ofconcepts important in investigations of systems^wholeness, sum, emergence,open and closed systems, the entropyof systems, the principleof equifinality,and so on.^' But the most puzzling feature emerges clearly in an essay by Bertal-anffy published in 1962. Bertalanffy distinguished between systems theory in'thebroad sense' and general system theory 'in the narrower sense'. He identified sevendifferent bodiesoftheory asconstituting systems theory in 'the broad sense',namely: cybernetics, information theory, game theory, decision theory, topology,factor analysis, and finally ,28. Hall and Fagen, 'DefinitionofSystem', particularly p. 8 2.29. See footnote330. For example, Kenneth E. Boulding, 'General Systems Theory^-The SkeletonofScience'in Buckley (ed.), Modern Systems Research.31. See Modern Systems Research PartsIIand III.32. Bertalanffy, 'General Systems TheoryA Critical Review' inModern Systems Research


10/14

CUS

The puzzle here is what these seven areas of systems theory have in common.Apart from the trivial fact that they all can have the word 'system' applied tothem, they seem to have little, if anything, in common. Certainly GST is differentin kind from the other six areas of theory mentioned by Bertalanffy; if these sixare regarded as second order studies, then GST is at an even greater level ofabstraction and must be classified as a third-order study.^^ It follows that it is aserious confusion to group these second- and third-order studies together as ifthey were analogous or parallel studies. Furthermore, the opposition of GST tothe mechanistic or analytic method, and its resemblances to Hegelianism and thebiological organieism of the late nineteenth century, again mark it off as quitedifferent from the other six areas of theory. If these basic tenets of GST wererejected, the other six bodies of theory identified by Bertalanffy apparently wouldbe able to stand virtually unshaken by the calamity that had overtaken GST. Orto put the point another way, studies or theories of a lower order are always ableto survive the demise of a theory of a higher order of abstraction, and so the sixsecond-order systems studies could survive the death of the third-order study,GST. It is this third-order study that is the centre of interest for the present paper.The objections brought against GST do not necessarily apply to these other areasof 'systems theory in the broad sense' .4. The weakness of the charges brought against the analytic or mechanistic

method ^The nature of the objections levelled by general system theorists against thetraditional analytic or mechanistic method has already been outlined. This methodis useful in many areas of science, they say, but it is not applicable to biology orany other area where complex systems are being studied. As Anatol Rapoportexpressed it:Biological processes are simply too complex to yield to the analytic method. . . .When we turn to attempts to subject human behavior to scientific analysis, theproblem becomes even more severe. . . . It follows that understanding cannot beextended beyond the scope of physical science without introducing concepts whichembody irreducible wholes in place of physically measurable variables.^ *

Or, in Bertalanffy's words:Thus the problem of wholeness and organization sets a limit to the analytical andsummative description and explanation.^'A further weakness of the analytic method, it has been claimed, is that it does not

33. According to the usage adopted here, a first-order study would be a study of 'particulars'^-e.g., a particular feedback system; a second-order study would be at the next level ofgeneralitye.g., a study of feedback systems in general; a third-order study would be evenmore generale.g., a study of many different types of system (feedback, information,biological, and so on).34. Foreword in Buckley, Modern Systems Research p. xvii.


11/14

SYSTEMS THEORY DIS REDITED PHILOSOPHY

do justice to thehierarchical organization found incom plex systems.^* T he pa r-ticular fact which the analytic method cannot deal with is that

at certain stagesofcomplexityin the interrelationsofcom ponents,a discontinuity. . . or emergent leveloforganization with novel features, may develop.^'Thisisthe w ell-known proble mof emergence. Traditionally ithas be en illustratedby the following example: A moleculeof thecom poun d, wa ter, con sists of tw oatomsof hydrogen and one atom of oxygen in combination, hence the formulaH a d . If these hydrogen andoxygen atoms were studied in isolation from eachother, and their prope rties determ ined, it could not be deduced from thisinformation thatifthey were comb ined they would form the colourless, odourlessand tasteless liquid we label with the name 'water' . The combinationofthe atoms,and the resultant production of a substance with unpredictab le pro perties, is anexampleof the phenomenonofem ergence.

These difficulties are not quiteasdifficult forthe suppo rtersofthe mech anisticor analytic methodtoanswerasgeneral system theorists seemtoimagine.In thefirst place, two conditions arenecessary for themechan istic explan ation of anysystem: the law or laws applicable to thesystem have to be known, and theinitial conditions havetobe known.^s Con sideras anexamplea system consistingof a container full ofwater, into one co rnerof which som e liquid dyehasbeenintroduced. With the passageof time the dye will diffuse throughout the container.Forasatisfactory e xplan ationofthis particular phenom enon to begiven, the lawof diffusion has to beknown, and so also doesthe set of initial conditions thenatureofthe dye, its original tem pera ture, and its original positioninthe containerof water, etc.). The same two general conditions are necessaryforthe explanationof more complex systems,buthereof course thediscovery of thelaws is moredifficult, and there aresoma ny factors involved that the statem ent of the initialconditions is anextremely com plex task. G ener al system theo rists, however, havenot shown thatinprinciple eitherofthe se tw o difficult task sis impossible; neitherhave they shown that thereisany effective sub stitu teforthe piecem eal investiga-tion of parts of the system the m ethod of investigation that is thecoreof themechanistic method. Bytheir own creed, systems theorists arebound to studythe wholewhich effectively is the wholeof reality, a nd this is a method whichin practice cannot be strictiy followed.As Bertrand Russell pointed out, if know-ledge is knowledge of the whole, then the attainment of knowledge becom esimpossible.

Secondly, thequestion of emergence hasoften been mishandled becauseoffailuretorecognizeacentral poin t. The p hilosopherofscience, Ern est Nage l,haswritten:36.For aninteresting discussion of hierarchical organization, and a novel vocabularytofacilitate discussion of this type of organization see Arthur Koestler, The Ghostin thMachine London 1967. In the book Koestler acknowledges his debt to Bertalanffy.37. Walter Buckley, Introduction to Part II in Buckley (ed.). odern SystemsResearch p. 37.


12/14

CUS

The logical point constituting the core of the doctrine of emergence is applicable toall areas of inquiry and is as relevant to the analysis of explanations within mechanicsand physics generally as it is to discussions of the laws of other sciences.''The logical point that Nagel refers to is that the conclusion of a valid deductioncannot contain an expression that does not appear in the premises; this indeed isthe core of the issue, because scientific explanations can be put in the form ofdeductions. It follows from this logical point that the possession of an emergentproperty cannot be deduced from premises that do not contain reference to thisproperty; it is logically impossible, for example, to deduce the production of acolourless, odourless and tasteless liquid (i.e. water), from premises that referonly to the properties of the gaseous substances hydrogen and oxygen. Thus,although the mechanistic method may fail to deal satisfactorily with emergentproperties, no other method could do any better, for the stumbling block is alogical impossibility.5 . The failure of GST as a scientific theory

Perhaps the most important feature of a scientific theory is its predictive valueWith the aid of a theory it should be possible to make a prediction about somefuture observable event. Sir Karl Popper has explicated this predictive functionof science in detail,^" and he has argued that non-scientific theories (or pseudo-scientific theories) do not have predictive value. In one interesting passage. Popperwrote of his own youth when four theories aroused great interest^Einstein'stheory of relativity, Marx's theory of history, Freud's psycho-analysis, and AlfredAdler 's individual psychology. From Einstein's theory an improbable-lookingprediction was made, and during an eclipse in 1919 Eddington made observationsthat could have refuted Einstein's theory; however, the theory stood the test andwas corroborated. The case was quite different with the other three theories:

I found that those of my friends who were admirers of Marx, Freud, and Adler, wereimpressed by a number of points common to these theories, and especially by theirapparent explanatory power These theories appeared to be able to explain practicallyeverything that happened within the fields to which they referred. . . . Once youreyes were thus opened you saw confirming instances everywhere: the world was fullof verifications of the theory . W hat ever happened always confirmed it. . . . It wasprecisely this factthat they always fitted, that they were always confirmedwhich in the eyes of their admirers constituted the strongest argument in favour ofthese theories. It began to dawn on me that this apparent strength was in fact theirweakness.*

In Popper's view, these theories were not scientific; they were not useful inmaking predictions, hence they never took a risk and they never faced even thepossibility of refutation.

It is interesting to consider GST in the light of Popper's remarks. Like Marxismor Freudianism, GST is wise after the event^it is always possible for a general39 . The Structure of Science pp. 372-3.40 . Karl Popper The Logic of Scientific Discovery London 1965.


13/14

SYSTEMS THEORY A DlS REDtTED PHILOSOPHYsystem theoristtoexplaininhis own theoretical terms what has happened in aparticular system he is investigating, but only afterthe 'happening' has taken place.The theorist's predictions beforetheevent usually are notpredictions in thescientific senseatall^they are far too vague. (The exception would be when thetheorist is investigatinga system where precise laws have been discovered by thetraditional mechanistic means, e.g., electronic circuits, mechanical systems,orcertain physiological systems.)*^

Asanexampleof the weaknessof systems theory with respect topredictivepowerand hence its weaknessas ascientific theoryconsider the recent book,The World ducationalCrisis: A Systems Analysis by Philip H. Coombs, Directorof the International Institute for Educational Planning,adivisionofUNESCO .*'Coombs studied the educational systemsofmany emerging and developed nations,his aim being to determine the effects on these systemsofchangesinvarious oftheir parts. He concentrated on such thingsasshortageof teachers, increasesincosts, and increases in the number of pupils; the great changes in these items overthe last few years constitutes the present world crisis in education. tisamarkedfeatureofCoombs' book, however, that he is unable to m ake precise predictions.From the fact thata given educational system has had a threefold increaseinincoming pupils ina very short period, for example. Coombsisunable to predicthow the system will adjust to this serious alterationofinput. tisa truism to saythat some changes must occur; furthermore,it ispossibletomakeaguessas tothe likely changes,for the rangeofpossibilitiesis a fairly limited one. Butinorder to make this sort of guess one does not have to bea systems theorist.This weaknessofGST seems to have been recognized by its keenest supporter.Perhapsit is well to end the paper whereit began, with Bertalanffy. n acriticalreviewofGSTin 1962, he wrote:The decisive questionisthatof the explanatory and predictive valueofthe newtheories attacking thehost of problems around who leness, teleology, etc. . . .There is noquestion that new horizons have been opened up but therelations toempiricat facts often remain tenuous.^

42. Also included here would bepredictions made by forexample, a cybemeticist or aninformations expert. As was pointed out earlier, these areasof specialization are not subjectto the serious criticisms that can be levelled against GST.43. New York 1968.44. 'General System TheoryA Critical Review' in Buckley (ed.). Modern Systems Researchp.21.


14/14

systems theory philosophy phillips 1969

Documents