LUND UNIVERSITY

PO Box 117221 00 Lund+46 46-222 00 00

What is systems science?

Löfgren, Lars

Published in:Cybernetics and systems 2002

2002

Link to publication

Citation for published version (APA):Löfgren, L. (2002). What is systems science? In R. Trappl (Ed.), Cybernetics and systems 2002 (Vol. 1, pp. 11-16). Austrian Society for Cybernetic Studies.

Total number of authors:1

General rightsUnless other specific re-use rights are stated the following general rights apply:Copyright and moral rights for the publications made accessible in the public portal are retained by the authorsand/or other copyright owners and it is a condition of accessing publications that users recognise and abide by thelegal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private studyor research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal

Read more about Creative commons licenses: https://creativecommons.org/licenses/Take down policyIf you believe that this document breaches copyright please contact us providing details, and we will removeaccess to the work immediately and investigate your claim.

WHAT IS SYSTEMS SCIENCE? LARS LÖFGREN University of Lund

Printed in: Cybernetics and Systems 2002, edited by Robert Trappl, Austrian Society of Cybernetic Studies, Vienna, Vol. 1, 2002, pp. 11-16. ISBN 3-85206-160-1

What is Systems Science?

Lars Lofgren (email: Lars.Lofgren@it.lth.se)

Systems Research, University of Lund, Box 118, S–221 00 Lund

[Printed in: Cybernetics and Systems ’02, Vol. 1, pp 11-16 (2002). Editor: Trappl, R,University of Vienna and Austrian Society for Cybernetic Studies]

Abstract

A most natural view is that systems sci-ence is a science dealing with systems andtheir systems properties. What, then, is sci-ence that it can deal with systems proper-ties which by nature are deeply penetratingalso into the systemic nature of our concep-tual processes, of language, and of scienceitself. We start out from an independentview of the concept of science. Namely, thatsciences are deductive in the sense that theresults of scientific activity are deductivelypresentable as in descriptive theories. Thisis what allows wide communication of scien-tific propositions, necessary for their acces-sibility, examination and tests by other sci-entists, eventually to be intersubjectively ac-cepted. On the other hand, the scientific ac-tivity is in general a process which is beyondfull deductive description. We examplify atendency. The more introspectively orientedthe domain of inquiry for a science is, themore difficult is it to isolate the deductive-result part of a science from the scientific ac-tivity. A systemic view of science arizes withscience referring not only to the deductive-result part of science but also to the scien-tific activity. We make it explicit that “sci-ence” in “systems science” be systemicallyconceived, and develop the concepts of sys-tems science, general systems, and system,accordingly.

1 What is science?

Sciences are deductive in the sense that the resultsof scientific activity are deductively presentable as indescriptive theories. This is what allows wide commu-nication of scientific propositions, necessary for theiraccessibility, examination and tests by other scientists,eventually to be intersubjectively accepted.

On the other hand, the scientific activity is in gen-eral a process which is beyond full deductive descrip-tion. In other words, scientific processes may well bebeyond scientific description. This is sometimes un-derstood with reference to the inductive nature of pro-ducing scientific hypotheses and a general agreementon induction as beyond deduction.

This activity/result distinction (or induction/de-duction distinction) for science is not always appre-ciated, however, which easily causes confusion as towhat science really is. It frequently happends thatscientific thinking becomes identified with deductivethinking – and science with deductive science. If so,the fundamental question of where the axioms (onwhich the deductive scientific processes build) comefrom, are neglected as “non-scientific” or, if consid-ered, dealt with in foundational research for the de-ductive science.

This situation prevails also for mathematics andlogics as deductive disciplines. In the following twoquotes (notably the last paragraph of the first), wesee how mathematicians (barely) may manage to avoidfoundational questions in trying to isolate mathemat-ics as a purely deductive discipline.

[Mostowski A, 1966, page 140] “The ab-stract set theory has contributed more thanany other branch of mathematics to the de-velopment of foundational studies. The rea-sons for this phenomenon are numerous.One of the basic assumptions of set theoryis the axiom of infinity which says that thereexist infinite sets. This assumption impliesthat the scale of infinte cardinals is itself in-finite. Thus the axiom of infinity leads us outof the mathematical domains which are closeto everyday practice and even to scientific ex-perience. We are thus faced at the very be-ginning of set theory with the fundamentalquestion of the philosophy of mathematics:which mathematical objects are admissibleand why? ...Most mathematicians do not perceive the

problem which is posed by the abstractnessof set theory. They prefer to take an aloofattitude and pretend not to be interested inphilosophical (as opposed to purely mathe-matical) questions. In practice this simplymeans that they limit themselves to deducingtheorems from axioms which were proposedto them by some authorities.”

[Whitehead and Russell, 1962, page v]“We have, however, avoided both contro-versy and general philosophy, and made ourstatements dogmatic in form. The justifica-tion for this is that the chief reason in favourof any theory on the principles of mathemat-ics must always be inductive, i.e., it mustlie in the fact that the theory in questionenables us to deduce ordinary mathematics.In mathematics, the greatest degree of self-evidence is usually not to be found quiteat the beginning, but at some later point;hence the early deductions, until they reachthis point, give reasons rather for believingthe premisses because true consequences fol-low from them, than for believing the con-sequences because they follow from the pre-misses.”

Concerning sciences, Russell is more outspoken onthe role of induction and the induction/deduction dis-tinction.

[Russell B, 1948, page 700] “What these ar-guments [referring to Hume] prove – and I donot think that the proof can be controverted– is, that induction is an independent logicalprinciple, incapable of being inferred [deduc-tively] either from experience or from otherlogical principles, and that without this prin-ciple science is impossible.”

It is interesting to see how Russell here (and nowhereelse, as far as we know) looks at induction as an in-dependent logical principle. Still, he is not alone. An-other similarly tolerant view is the following of vanBenthem:

[Van Benthem J, 1982, page 435] “Logic Itake to be the study of reasoning, whereverand however it occurs. Thus, in principle, anideal logician is interested both in that ac-tivity and its products, both in its normativeand its descriptive aspects, both in inductiveand deductive argument. ...An enligthened logician like Beth, for in-stance, realized the danger of intellectualsterility in a standard gambit like separatingthe genesis of knowledge in advance from itsjustification ...

[page 450] ... theories as scientific activitiesrather than products of such activities are notirrevocably outside the scope of logic. ...In the semantic perspective too, there isroom for pragmatic studies. E.g., model the-ory presupposes that successful interpreta-tion has taken place already. How? ...... these references are only the first land-marks in a hopefully fruitful new area oflogic.”

It is our definite impression that logics, as a de-ductive discipline, has today received such a generalacceptance, that it is far more advisable, and natural,to leave it as such – and, instead, broaden the conceptof language to have it include the inductive descriptionand interpretation processes.

Whether science is conceived deductively or systemi-cally (not in isolation from its foundations or genesis),it is bound to have its results presented deductively– for reasons of communicability. Communicabilityis necessary for intersubjective acceptance of scien-tific “truths” (and their presuppositions – if revealed).Communicability, in turn, presupposes language. Howis this concept to be conceived?

2 What is language?

With respect to common understandings, languagehas fared less well than logics. Let us recall someselected historical notes from logics and philosophy ofscience leading to our systemic conception of language.

In a disciplinary account of logic, as in mathemati-cal logic, the concept of language is either not definedat all, or is considered as partly outside the domain ofthe discipline. Compare Shoenfield’s book on mathe-matical logic:

[Shoenfield J, 1967, page 4] “We considera language to be completely specified whenits symbols and formulas are specified. Thismakes a language a purely syntactical ob-ject. Of course, most of our languages willhave a meaning (or several meanings); butthe meaning is not considered to be part ofthe language.”

Shoenfield’s honest account of his disciplinary ap-proach indicates how the fragmentation into math-ematical logic “makes” language devoid of meaning.This is a clearly distortive approach, or a high priceto be paid for mathematical clarity.

Also in a wider, philosophical linguistic context, ad-mitting meaning as part of language, the fragmenta-tion problem is apparent:

[Putnam H, 1975, page 215-216] “Analysisof the deep structure of linguistic forms gives

us an incomparably more powerful descrip-tion of the syntax of natural languages thanwe have ever had before. But the dimensionof language associated with the word ‘mean-ing’ is, in spite of the usual spate of heroic ifmisguided attempts, as much in the dark asit ever was.”...In my opinion, the reason that so-called se-mantics is in so much worse condition thansyntactic theory is that the prescientific con-cept on which semantics is based – the presci-entific concept of meaning – is itself in muchworse shape than the prescientific concept ofsyntax.”

In semiotics, sometimes referred to as the science oflanguage, there is an explicit recognition of syntax andsemantics, as well as pragmatics. In [Carnap, R, 1942,page 9] it is proposed that the three parts, syntax, se-mantics, pragmatics, constituting the whole science oflanguage, can be individually understod. In [LofgrenL, 2000, page 18] we argue that such a fragmentationis untenable. The argument is based on language be-ing a holistic (genuinely systemic) conception. It hasto be complementaristically comprehended.

Language, in its general systemic con-ception, is a whole of complementarydescription-interpretation processes. Com-plementarity refers to holistic situationswhere (a classical) fragmentation into partsdoes not succeed. In its complementaristicunderstanding, the phenomenon of languageis such a whole of description and interpre-tation processes, yet a whole which has nosuch parts fully expressible within the lan-guage itself. Instead, within the language,the parts are complementary (entangled) ortensioned. There are various related ways oflooking at this linguistic complementarity:

(i) as descriptional incompleteness: in nolanguage can its interpretation processbe completely described in the languageitself;

(ii) as a tension between describability andinterpretability within a language: in-creased describability implies decreasedinterpretability, and conversely;

(iii) as degrees of partiality of self–reference(introspection) within a language: com-plete self–reference within a language isimpossible;

(iv) as a principle of “nondetachability oflanguage”.

By comparison with Carnap’s fragmentation of lan-guage, we thus understand language as a whole

by complementaristic comprehension. The syntaxand semantics parts correspond to the complemen-tary description-interpretation processes, and prag-matics to the processual nature of the description-interpretation processes. However, while Carnapthinks of a classical fragmentation in the three parts(syntax, semantics, pragmatics), we recognize a nec-essary entanglement of them.

Languages may change and evolve, and with themtheir capacities for describing and interpreting. Yet,at each time that we want to communicate our actualknowledge, even on the evolution of language, we arein a linguistic predicament, namely to be confined toa language with its inescapable complementarity.

As explained in [Lofgren L, 1992] we have arguedthe validity of the linguistic complementarity from thefunctional role of any language, namely to admit com-munication or control.

Presupposition for language. Descrip-tions (sentences, theories) are always finitelyrepresentable (in order to be transmittablein a terminating communication) and lo-cally independent of time (remaining fixedfor as long as the description is being usedas description, i.e., is being transmitted incommunication, is being analyzed and inter-preted).

3 On von Bertalanffy’s unificationgoal for general systems

The concept of general system was conceived by vonBertalanffy in the hope of coming to grips with thedisciplinary fragmentation of science.

[von Bertalanffy L, 1968, page 30]. “TheQuest for a General System Theory. Modernscience is characterized by its ever-increasingspecialization, necessitated by the enormousamount of data, the complexity of techniquesand of theoretical structures within everyfield. Thus science is split into innumerabledisciplines continually generating new sub-disciplines. ..It is necessary to study not only parts andprocesses in isolation [as in classical analyticscientific methods], but to solve the decisiveproblems found in the organization and orderunifying them, resulting from dynamic inter-action of parts, and making the behavior ofparts different when studied in isolation orwithin the whole.”

Although von Bertalanffy apparently had sweepingideas of concepts like theory and science, his intuitionswere remarkably foresighting. Notably concerning thedifficulties of having “systems science” develop in away not itself subject to fragmentation.

[von Bertalanffy L, 1968, pages vii-viii].“The student in ’systems science’ receives atechnical training which makes systems the-ory – originally intended to overcome cur-rent overspecialization – into another of thehundreds of academic specialities. More-over, systems science, centered in computertechnology, cybernetics, automation and sys-tems engineering, appears to make the sys-tems idea another – and indeed the ultimate– technique to shape man and society evermore into the ’megamachine’ which Mum-ford (1967) has so impressively described inits advance through history.”

Under a section General System Theory in Education:The Production of Scientific Generalists he writes:

[von Bertalanffy L, 1968, page 49]. “Con-ventional education in physics, biology, psy-chology or the social sciences treats themas separate domains, the general trend be-ing that increasingly smaller subdomains be-come separate sciences, and this process isrepeated to the point where each special-ity becomes a triflingly small field, uncon-nected with the rest. In contrast, the edu-cational demand of training ’Scientific Gen-eralists’ and of developing interdisciplinary’basic principles’ are precisely those generalsystem theory tries to fill. They are not amere program or a pious wish since, as wehave tried to show, such theoretical structureis already in the process of development. Inthis sense, general system theory seems tobe an important headway towards interdisci-plinary synthesis and integrated education.”

In our view, the question is if “scientific generalists”can at all be trained in a university curriculum – wherecommunicational demands require communication tobe centered around deductively presentable results,necessarily leaving behind large parts of the (induc-tive) activities leading to the results.

The problem is the same as if foundational researchcould be trained in an orderly fashion. In our view,foundational insights could be expected first at seniorlevels, after having acquired disciplinary knowledge.

By way of another example of von Bertalanffy’s sys-temic insights, let us quote as follows.

[von Bertalanffy L, 1968, page 238]. “Itmay be mentioned, in passing, that the rela-tion between language and world view is notunidirectional but reciprocal, a fact whichwas not made sufficiently clear by Whorf.The structure of language seems to deter-mine which traits of reality are abstracted

and hence what form the categories of think-ing take on. On the other hand, the worldoutlook determines and forms the language.

A quite plausible observation. In particular we noticevon Bertalanffy’s broad conception of language. Yet,apparently not developed all the way to its systemicconception, and not recognizing its ultimate, unifyingpresupposition (cf section 2).

4 On Klir’s view of systems science

In his writings on systems science (by way of a smallsample, [Klir G, 1991; Klir G, 2001]), Klir starts froma common-sense definition of system, whereby:

“the term system stands for a set of thingsand a relation among the things. Formally,S = (T, R), where S, T, R denote respec-tively a system, a set of things, and a rela-tion (or, possibly, a set of relations) definedon T .”

Klir points at the simplicity of this definition as itsweakness as well as its strength.

“The definition is weak because it is too gen-eral and, consequently, of little pragmaticvalue. It is strong because it encompassesall other, more specific definitions of systems.Due to its full generality, the common-sensedefinition qualifies for a criterion by whichwe can determine whether any give object isa system or not: an object is a system if andonly if it can be described in the form thatconforms to [S = (T, R)].Once we have the capability of distinguishingobjects that are systems from those that arenot, it is natural to define systems scienceas the science whose objects of study are sys-tems.”

Consider the stated presupposition for the proposedsystems science definition, namely that “we have thecapability of distinguishing objects that are systemsfrom those that are not”. What if we don’t have such acapability? Shouldn’t this problem then be within thedomain of systems science, i.e., be a systems scienceproblem!

Compare Cantor’s common-sense conception of aset as a collection of definite, distinguishable objectsof our intuition or of our intellect to be conceived asa whole – referring as it obviously does to a personconceiving the whole in his inner cerebral languageand requiring translatability to an external communi-cation language in order to secure intersubjective ac-ceptance. Compare how the latter requirement leadsinto deep foundational studies which gradually ma-tures in actual set-theories (whose axioms cannot be

fully understood if isolated from the foundational do-main; compare the Mostowski-quote in section 1).

Compare computer science and the problem of iden-tifying the computable functions. This identificationcannot be done by computers alone, but requires foun-dational aids – which are often included in the conceptcomputer science (thus a systemically conceived sci-ence).

Also in quantum mechanics, as a physical science,we see hints at a systemic conception. From [Busch,Lahti, and Mittelstaedt, 1996] we quote:

“The quantum theory of measurement is mo-tivated by the idea of the universal validityof quantum mechanics, according to whichthis theory should be applicable, in partic-ular, to the measurement process. Henceone would expect, and most researchers inthe foundations of quantum mechanics havedone so, that the problem of measurementshould be solvable within quantum mechan-ics. The long history of this problem showsthat, in spite of many important partial re-sults, there seems to be no straightforwardroute towards its solution. This general im-pression is confirmed in the present work bymeans of a number of no-go-theorems.”

We have not seen Klir embark on foundational ques-tions for justification of his above systems science pre-supposition. We do not think that his concept of sci-ence, in systems science, is systemically conceived, andnot that his concept of system is wide enough to per-mit natural introspective system-inquiries.

5 Systems Science, General Systems,and System – systemically conceived.

With reference to the systemic nature of our linguisticcomprehension processes (sections 1 and 2), we arguethe basic systems-concepts as follows.

Systems Science is to be systemically conceived.That is, systems science should not (and could notin general) be isolated as in referring only to the de-ductively presentable results of systems research – butshould be allowed also to refer (in an inductive vocab-ulary) to the very activity of systems research.

This, in general, requires a shift from theory, logic,science – conceived as deductively presentable disci-plines – to systemic language as the natural back-ground for systems research as a foundational activity.

General Systems, focusing on the property of be-ing systemic (paradigmatically present in language),may well be identified with systemic language (as awhole of complementary, or entangled, description-interpretation processes).

System, finally, is (as usual) regarded as a whole ofinteracting components (parts) – with the (not usually

stated) qualification: conceivable as such in a sharedlanguage.

Without the last qualification, we may be in dangerof violating the principle of “nondetachability of lan-guage” (view iv of the linguistic complementarity) –easily causing difficulties in introspective systems con-texts.

6 Hints at applications

We have argued a shift in attention, from the deduc-tive scientific disciplines to the systemic concept oflanguage, in order to come to grips with problemsinherent in a one-sided classical focus on the deduc-tively presentable results of scientific activity (as iffragmentable from the activity itself). Let us hint atsome applications of this systems move.

6.1 Unification of fragmented disciplinesAs mentioned in section 3, von Bertalanffy conceivedof his concept “general systems” in the hope of comingto grips with the disciplinary fragmentation of science.At the same time he was aware of a fragmentary de-velopment of general systems itself.

Moving to the systemic concept of language asparadigm for general systems (section 5), we noticethat it is unifying, first in the sense that it obtainsfor several species of language, from genetic language,over formal languages, to external communication lan-guages. Further, our actual shared communicationlanguage is unifying in that all communicable knowl-edge depends on it. Knowledge in various scientificdisciplines, even if in distinct languages, is unified inthe common presupposition of the linguistic comple-mentarity (section 2).

Let us expand on systems research, as foundationalactivity (section 5), as unifying. Foundational re-search, as search for founding scientific truths, often it-self leads to fragmented foundational disciplines. Thisis due to the contextuality of the semantic truth con-cept.

However, there is another approach to foundationalstudies, namely in a succesive revelation of presuppo-sitions, intimately tied to view i of the linguistic com-plementarity (see [Lofgren L, 2002]). This is unifyingin that it ultimately will lead to the presupposition forthe linguistic complementarity.

Beside unifying understandings of fragmented sci-entific disciplines, we have unification in interdisci-plinary domains, notably in systems engineering. Forexample, in the Hubble telescope system, various sci-ences meet in its construction.

6.2 Godel’s revision of Formal SystemAttempting to answer the question what a system is,[Rosen R, 1986] compares the concept of system withthat of set. One of his comments is: “since the axiomsystems in terms of which set-ness is characterized are

themselves systems, it may well be that attempts todefine systemhood in terms of set-ness is cart beforehorse”.

Of particular interest here is that Godel, in his fun-damental papers from the early thirties defined theconcept of formal system (the axiom system to whichRosen refers) by means of a concept of “finite proce-dure”. And that Godel in 1965 made a revision of hisconcept of formal system, namely by replacing “finiteprocedure” by that of a Turing machine. As explainedin [Lofgren L, 1992], we look at this as a first embry-onic step from formal system to systemic language (afull step also recognizing the Turing machine as inter-pretor in a programming language).

In this perspective, Rosen’s “cart before horsedilemma” resolves nicely in our conception of system(section 5; with its stated relation to language).

6.3 Linguistic realismIn [Lofgren L, 1977; Lofgren L, 1993] we explain theconcept of existence in linguistic realism by referringto the very nature of an inductive linguistic confirma-tion process. This has vast explicatory consequences.

For example, for understanding the fragmentationproblem: is nature in itself fragmentable, and therebynondistortively fragmentable, or, is it our linguisticdescription processes which make nature appear frag-mentable?

Again, for understanding the “intended interpreta-tions” problem, with its sources in the Lowenheim-Skolem theorem, central within current inquiries into“the limits of logic” – and possible resolutions af-ter taking the step into a shared systemic language([Lofgren L, 2002].

References

[Van Benthem J, 1982] “The Logical Study of Sci-ence.” Synthese, 51, 431-472.

[von Bertalanffy L, 1968] General System Theory.New York: George Braziller.

[Busch, Lahti, and Mittelstaedt, 1996] The QuantumTheory of Measurement. Berlin: Springer.

[Carnap, R, 1942] Introduction to Semantics and For-malization of Logic. Cambridge, Massachusetts:Harward University Press.

[Klir G, 1991] Facets of Systems Science, New York:Plenum Press.

[Klir G, 2001] “Systems Science” (Manuscript)

[Lofgren L, 1977] “On Existence and Existential Per-ception”. Synthese, 35, 431-445.

[Lofgren L, 1992] “Complementarity in Language;Toward a General Understanding.” Pp 113-153

in Carvallo M, ed., Nature, Cognition and Sys-tem II. Dordrecht: Kluwer.

[Lofgren L, 1993] “Linguistic Realism and Issues inQuantum Philosophy”. In Laurikainen K, Mon-tonen C, eds., Symposia on the Foundations ofModern Physics 1992: the Copenhagen Inter-pretation and Wolfgang Pauli : Helsinki, Fin-land, June–August 1992. London: World Scien-tific, 297–318.

[Lofgren L, 2000] “Fragmentability, a FoundationalSystems Concept.” Pp 15-20 in Trappl R, ed.,Cybernetics and Systems 2000, Vol 1. AustrianSociety for Cybernetic Studies.

[Lofgren L, 2002] “Unifying Foundations – to be Seenin the Phenomenon of Language.” Forthcomingin Foundations of Science.

[Mostowski A, 1966] Thirty Years of FoundationalStudies : Lectures on the Development of Math-ematical Logic and the Study of the Foundationsof Mathematics in 1930 – 1964. Oxford: BasilBlackwell.

[Putnam H, 1975] Mind, Language and Reality:Philosophical Papers, Volume 2. Cambridge:Cambridge University Press.

[Rosen R, 1986] “Some Comments on Systems andSystems Theory”. Int J General Systems, 13, 1-3.

[Russell B, 1948] History of Western Philosophy.London: George Allen and Unwin.

[Shoenfield J, 1967] Mathematical Logic. Reading,Mass.: Addison-Wesley.

[Whitehead and Russell, 1962] Principia Mathe-matica. Cambridge: Cambridge University Press.First edition in 1910.