rNFORMA nON PROCESSING 80, S.H. LavUlgton (ed.) Nonh·HoUaJ!d Publishing Company © LFIP.1980

lnviled Paper

, THE LIMITS OF TOGETHERNESS · System Research Ltd., Woodville House, 37 Sheen Road Richmond. SUrrey. UK

This paper makes a distinction between conversation and communication (signal transfer which may, or may not, be conversational) . The word " conversation" is given an interpretation, which ref.1nes its COIIBDonsense meaning. conversation maintains the autonor;Jy or identity of systems and, also, generates independencies between systems (human , societal, or others), which is a prerequisite of dialogue.

The natural habitat of human beings is, increasingly, an "information environment" where communication and computation have altered suppositions about signal distance, or "togetherness". Limits are discussed with emph... sis upon the hazards engendered by too much togetherness in person-person or wan-machine interaction. Because it is possible to comment, cogently, upon the nature of conversation. these limits can be recognised and remedie's to the malfunctioning which is likely to occur if they are exceeded, are proposed .

The Limits of togetherness The technical aspect of the paper is devoted to saying what Conversation is . As a prelimi- The popularised word "togetherness" aptly nary definition, Conversation is "Concept captures a general notion of human proximity, sharing" . of meeting and speaking, or dancing together at a festival. Social groups. be they families 1 . 1 Logical Discrimination. urban communities Or the older universities, have institutions which promote togetherness; The value of a communication is accuracy and the dining table, a market, or a cafe 3S the veridicality imaged by the a nd case may be. On more or less ritualised values of propositional or descriptive logic occ3sions, and in the traditional places, hu- (or its prob3bilistic extrapolation) where mans converse; either verbally, or by image "True" or " False" are usually modelled as and gesture. I submit that the oonversation states 1, O. In contrast, the value of a which occurs, debate and sometimes agreement , conversation is agreement, reached by is the stuff of ciVilised life and together- commanding and obeying or questioning and ness is essential to it. On the other hand answering (neither questions nor commands have ther e are also limits upon "togetherne ss"; too factual truth values). Agreement may be imaged much of it, for example. gives rise to by the coherence truth values of a procedural specific symptoms of individual and social logic and these values are conveniently malaise. These symptoms typically appear when modelled by phySical coherence between non- the communication, allowed by proximity. is linear processes . not conversation. In communication. inform4t!on transfer is 1 . Introduction founded upon selection, albeit statistical, 3mOngst states of t ransmitters and receivers Communication and conversation are distinct, that are extra-theoretically specified as and they do not always go hand in hand. Suppose independent. (apart from the cOllllllUnication tha t communication is liberally construed as channels) but synchronised . for example, by a the t r ansmission and transformation of signals. recognisible punctuating symbol . Notions such If so, conversation requires at least some as "noise" and "capacity" rest upon t hese communication . But , enigmatically perhaps, very foundations. In contrast, the information bad communication may admit very good conver- transfer of a conversation is of the Petri- sation and the existence of a perfect channel Bolt [1] t he extent to which other- is no guarantee that any conversation will take wise independent participants are rendered place. locally dependent , Or otherwise asynchronous participants become locally synchro nised when Bec3use communicatio n theory is well known, agreements are reached, as a result of concept differences between communication and conver- sharing. Unlike transmitters and receivers, sation can be pointed out by comparison and t he participants. who converse dOd share contrast, at this juncture. concepts, are not unambiguously predefined and are not extra-theoretical importat ions. The act of conversing surely depends upon their autonomy or distinction but, also. this act .. System Re8earon. Ltd. generates a distinction . and Brun.eZ Univer8ity, Engtand. Netn.e1'Iands fop Advanced Studies, Wassenaar, Netn.erlands.

999 1000 G. Pask/The limits of togetherness , •

1.2 Some Background 1.3 An Information Environment • • My own conviction about the importance of In the past, conversation has often been conversation and its critical relation with hampered by lack of communication . In the t ogetherness took shape slowly. During the future, the familiar barriers, such as geo- later 19505 and early 60s, We constructed what graphical distance, are unlikely to be obtru- were glibly called "adaptive teaching machines'! sive; conversation will be more endangered by , These devices operated in Ski lls as varied as excessive togetherness; the possibility of typewriting, track ing and problem solvi ng communication can be safely assumed to exis t. [ 2, 3, 4, 5, 6, 7, 8, 9 ]. The matter is especially significant i n the context of well known developments in communi- I n this context the word "machine" means a cation, data storage, and (classical) computa- piece of hardware constrained, algebraically, tion, which are rapidly creating an "informa- to act as a computer; nowadays ,for example, the tion environment". systems are more reliably implemented using standard microprocessors . 'Ihe word "teaching" There are,first of all, technological develop- suggests that someone engages in conversation, ments. For example, in communication, fibre usually non-verbal, but with the "te aching optic technol ogy, provides virtually unlimited machine" , bandwidth channels in urban districts, microwave and satellite communication over larger dis- Only in the 1960s and early 1970s was it clear tances, in stor age, video discs,high capacity that this suggestion is utter nonsense. The ( > 80 megabyte) magnetic discs , and the algebrai c constraints whi ch give "machinehood" complementa-ry main store techniques of low to the hardwar e are designed to prohibit energy semiconductors and magnetic domain conversation with machines . For instance, you (bubble) devices; in computation, widely cannot, by definition, "disagre e" with a disseminated microprocessors to connect storage machine; you can onl y say it is a "brOken" media to channels and provide local processing machine, Of cour se, people may and do capability; optical and array processors. communicate with machines (which, if they Others, amongst them Hines [13J, have argued, belong to this category are better called cogently and convicingly, for these technolo- "training machines") . Also, given a different gies and the soci al". national and industrial design, such as the learning monitor CASTE pressures that are, willy nilly, going to carry [ 10, 11, 12] Conversation may take place technical breakthroughs int o t angible struc- through (not with) machines, which exteriorise tures . the concepts that are shared as tangible re- cords , As a result, it was possible to obtain Nex t, there are developments,for the most part empirical support for a theory of such trans- anticipating the currently burgeoning technolO- actions and, going a little further, to con- gical base, concerned with information handling struct hardware artifacts which do have conver- and processing. For examfle there are systems s ational capability because they-are not, for- like Nelson's Hypertext 14]Winograd and Kaye's mally speaking, machines. (Conversation with [15J KRL. There is Negroponte's "data space", machines is disallowe d for i!llgehraic reasons, and the video disc store of Aspen, Colorado, s o not to do with material embodiment . Brains are accessed that a user can drive through the biological after all). streets, industries, history or everydayacti- vity of that town l16J. My OWn entail ment These man/machine studies were set in the meshes, perhi!lps, are candidates -[171. There is a context of learning, hypothe sis construction, host of rea ctive animated graphics facilities , and various k i nds of design. At the moment, the exemplified by De Fantis' work. chief foc us is upon complex team decision making, which includes _planning and strategy These developments, combined with the technical generation. The work provides a lar ge, but advances and the pressures to implement them, -• laboratory contrived, sit uation in which the lend substance to the claim that communication/ togetherness of people (or even the perpectives computation proximity is no longer just a matter and roles they adopt) is determined by a - of geography. Rather, the natural environment , communication/computation medium in which of mankind becomes increasingly an information some, but only some, of the transactions are environment, chiefly determined by these conversational. It seem'that this miniature communication/computation systems. This c l aim is world is a long way from a real world of not confined to dense conurbations (as it might geographical distance and transportation have been a few years ago), nor is it a claim delays, the traditional indices of about the unforseeable future. It is a simple extrapolation from currently available facts There is, of course, a great difference in and figures. scal e , but I claim (a) that the real world i5 changing, due to various trends and technolo- Forseeably, the industrial, social and national gies, so that its form is strictly comparable pressures which promote information technology • to these laboratory situations; (b) that the will also give riSe to legisl a t ion against maladies and misfunctions due to a failure for privacy o r isolation, all quit e jus tifiable and one reason Or another to converse,are typifiec "for our Own good". Before long, the statute by mechanisms observable, in the miniature, books will ordain that any partition, wall or and (c) that remedies which p r omote conver sa- enclosure is penetrated by a channel o f so many t ion in the miniature situation are likely megahertz capacity, and will inClude rules like to p r ove e f fective in the r eal world of the "a f ibre- optic-cable shall connect each legal futur e . house" . • • G. Pask/The limits of to get he mess 1001

• • Little is known about the ecology of an infor- Robinson [23] [24] amongst many others) that mation environment where distance is 5igna1- viable medium-to-large corporations, schools distance conjoined .to localised storage and of thought (maybe committees) are "supr3human" computer it is fair to suppose that organisms which have, in a very real sense, an the pathOlogies manifest today at the limit s autonomy of their own. We communicate with them, • of togetherness on the occasions when there is by posting memoranda, receiving edicts, fiats, communication which looks like conversation but etc., but do not converse (s.hare c;on,"-epts) with is not at all conversational, will be amplified them. Again, the boundaries are not clearcut; .tn an information environment. They appear as we do converse with our families, our extended major hazards in the future. families, members of a club, personal friends . But can one, for example, converse with the 1.4 Illustrations entire colloquy of professional peers ? That largely depends upon whether we have learned to The following maxims e x emplify the pathologies translate natural-language concepts into the in question. (a) "Communication need not be esoteric-language concepts, natural for organs conversation". For example, committees are of the body or diviSions of an industrial often said to decide as a result of debate enterprise. (conversat ion) between their members. By of ·debates the committee members agree, or they (c) "Too much togetherness inhibits conversa- agree to be at loggerheads . On salient topics, tion". This maxim is illustrated by life admidst however a coherent view is reached. This ideal open-plan architecture, in vogue some years ago; is seldom approximated, at any rate amongst apartments and houses where (as a phrase of that the committees that proliferate in academia. epoch)) "parents grew up with their children". Of course, a committee plays a social role; Under these circumstances conversation is im- it gives reason for the members to gather and paired. If people live in such indecent proxi- it provides a valuable forum for rhetoric. But, mity they cannot easily sustain the autonomy whaterer else, big committees do not themselves of participants, who might converse together, decide. On the contrary, the communication of having distinct perpectives and pOints of "business in hand" absorbs, rather than pro- motes, debate (Atkin [18J). In practice, deci- (d) "When there is too much togetherness s ions are made by persons, or small groups who communication acts as a mechanism of isolation do converse. The larger consensus amounts to rather than a vehicle for dialogue". For distr.i,bution of blame; a "committee decision", example, in my culture, dinn er is an occasion for which no one is responsible. for di9 cussion. It is frightening to see a family who gaze at a television set over their (b) "Both subhuman and suprahuman organisations evening meal, althQugh the behaviour is typical communicate with humans, but do not converse of open'-plan living. Clear ly, a simultaneous with them" . It is a truism that any organism, conversation is out of court. Again clearly, such as a human being , depends upon communica- the televisual communication serves to isolate tion with, and amongst, organs o f t he body the diners and give them the autonomy they need and resources, for example, of food . AS humans in order to be people (but also, the communi- we call these necessary components and resour- cation of TV prevents the use of that autonomy ces "subhuman". We may credit them with life for conversing with each other). but do not, as a rule, converse with them . The bounds are not entirely clearcut. For instance, (e) "Too little apPflrent-togetherness promotes after practice (perhaps, a i ded by biofeedback) unil'orm-surrogate-tog'etherness". Suburbs you may, in a very real sense, l earn to exemplify thiS dictum. If you live 20 miles conditionally regulat e your heart beat and from your office, then you commute in a uniform rationally influence a functional system which pattern, See and hear the mass- media, are part usually operates automatically. Contrariwise of the market for microprocessor games, and, there is ample evidence (Beer [2 1J, Robb [22J l ikely as not, your youngs t ers play games identical to their geographically dispersed , • neighbours. Amongst the mechanisms promoting uniformity, temporal synchronisation whether of rush hours, or viewing peaks, produces a covert togetherness devoid of conversation •

1 .5 Commentary •Eclectically minded anthropologists and ecol- These examples of potential pathology been ogists p01:nt ou t that if man is to sU"('1)iVe he chosen because they are poignant, generally must learn to live with nature. Bateson [l9J relevant to such issues as computer conferen- and Illieh [20], for example, support thi s cing (it could be a boon, or it could destroy But view in very different ways . both of them the social raison d'etre of committees), and are auxu>e that naive images of "returning to universally available microprocessors . They are nature" have no more than wcal relevance (the also snapshots of amplifying, self- replicating, can loealities in question be quite large, of and self-stabilising processes that grow and coupse), and that renders communication! stereotype by entirely systemic mechaniSms. computation essential . In some conditions the mechanisms are intellec- tual, in others, the mechanisms a re concrete. For instance, the "togetherness-movement" g,,-ve rise to open p lan structures which were, in turn, adopted by property developers as cost .. 1002 C. Pask/The limits oflogelhemess

e ffective . In turn , again, the existence of entity may be quite alien, even if they are open-plan a ccommodation induced a social pheno- concepts of an abstract kind, like "circle" or menon of "open -plan living" . "rectangle" , which have standard tex tbook de- finitions . Because the mechanisms are systemic, it is reasonable to suppose that they will operate in So, if l\ and B are asked about " circle", they • a n information environment, wh ere the probably come up with d ifferent explanations, constraints imposed by temporal or geographical for example, A draws a circle with a compass on , boundaries are of decreasing COnsequence. paper, B slices a cylinder. Equivalent ly, they , computation substantially eliminates the would write different computer programs in t emporal dimension and signal neighbour hood order to generate circles. Even if A and B know bends the spatial dimension . System d esigners the standard definit ion , "locus of a ll points , will be responsible for the kind of communica- equidistant from a given point , on the plane" tion and computation that goes on , a nd for (for inst ance, because they are s tudents in the whether or not conversation may take place . same geometry course) , the definition is They will be the engineers of togetherness, usually for examinations . and must respect its limits. A and B Can share concepts that are stable and 2. Main Theme g1ve.n names T , T , in a language ( s ay) L. B Conversely, tJ1:e common concept, t he "meaning of The ma i n conte ntion of this essay is that most the word", exists because of a concensus amongst currently available theories of communication a whOle community of L users, A, B, .. . N . and computation are not adequate tools for Concepts are shared by means of requests, COl'lI- e ngineering togetherness. Although these mands, persuasions, etc., from A, obeyed by B, theories are beautifully developed and have or vice versa; through questions posed by A and mathemat ical e legance, they are unable, without answered by S, or vice versa . This activity , extra- theoretic props , t o distinguish between whether it is verbal or in a language for communication and conversation . InSOfar as eXchanging, executing and debugging programs pathologies arise when conmunication looks (or a language of graphics, gesture, etc) !!. a like conversation but is" not conversationa], conversation . it res ults in a sharing of t hese theories do not provide a framework in some o r a ll of A' s, B'S concept, then there is which the pathol09ical lim! ts an agreement TAB between A and B, or, in o f t ogetherness can be detected. an agreeme nt , t he meaning of which for A, B, ... N is a · concept where "T" Certainly there are e xceptions which do not (omitting s ubscripts) is the word, i n the warrant this stricture (the work of Braten language, L, of this cmmnunity, that designates [25] , Byshovsky [26] , Flores and Winograd it. [27] , Gaines 29] , Gergely [30, 31] , Goguen,Varela [ 35] , Maturana [36J , McCulloch There is a danger of confusion between II static [37] and Von Foerster [38, 39 , 40 , 41, 42] ) . symbol , lik e T, and the philosopher ' s "meaning In order to give technical substance to the of T ". In the company of psychologist s, it is diSCUSSion, one of these theories, due to my easy to believe that a concept is a s tored own group, will be outlined . It is calle d templet, or pattern, in A' s or B's for "Conversation Theory" (abbreviated hence that is ofte n suggested . So far as this paper forward , to c or ) . [ 11 , 12 , 43 , 44, 45 , 46, 47, is concerned, the suggestion is outright denied; 48, 49 . SO] " concepts, in either case, are kinetic . Speci- fically, concepts are near- coherent bundles of CT is a ref lective (or participant) theory and procedures (that is, programs composed in a comments upon the "concept sharing" of language, L, and compiled for execution in a "participant s" (A, B, •.. ), upon their agreement brain Or some collection of brains) . Personal and failure to agree. CT is also a r elativistic concepts are executed to produce personal theory . The dat a structu res rel ative to which behaviours (like riding a bicycle) , which im- it is specified (entailment meshes), have a plicate A's or B's environment, or else they l ogic of distinction and coherence. They are are execut ed only in A's or B' s brain to pro- the "environme nt" or "domain" of conver sations duce descriptions, such as A's or B' s imagina- a nd offer an index of p r ox imity which is tion of riding a bicycle . Concepts determine interpretable a s together ne ss . skilled behaviour and, in the sense that A o r B have skills, concepts are skills, and, for Although terms like " concept" or "stable instance when thinking about geometry, are concept" have a perfectly straightforward purely i ntellectual skills. connotation, t hey are also used, with some re- finement o f meaning, as technical terms. 2.2 Stable Concepts

2. 1 _Concepts "M omorable" or " resilient" concepts (in contrast, to the evanescent trace of a phon! number wh i lst Philosophers s ay that a concept is the meaning it is dialled), are stable because they a r e of a word, (or any perceptible symbol , it need productive and reproduced (learned and relear- not be writ t e n or spoken) . Psychologis ts ascri- ned) . Without unique commitment to mental be concepts' t o people and speak of A' S concept operations, the criterion of stability equiva- of a hOllse , (or riding a bicycle or A's great- lent to "productive and r eproduced", is uncle) as distinct from B's concept of a house (or riding a bicycle, or B' s great uncle) . A' s lind B' S concept s of a cOWll\only name<'! e ve nt or

• • G. Pask/TIle limits of togetherness 1003

, "organisational closure"!! , or "autonomy"; of (themselves concepts), then (1) the original which the classical stabilities are special procedures are reproduced, and (2) fresh means cases. The criterion is in line with Von (procedures) for achieving the same end are N",umann's (51), Burke ' s (52) or Lofgren's (53, enerated and rendered • eventuall , coherent 54) work on self reproducing automata, and WIth other members of the collection (the near- Fig 1 is an outline of it. Concepts are prod- coherent-cluster'; of procedures, some or all of uctive and reproduced (relearned, reconstruct- which may be executed simultaneously). That is, ed) if, when operated upon by operations "some process is executed" and "collections of processes tend to (but do not necessarily reach) coherence" . Execution (Ex) of Productive and Reproductive Notice that the structure of an organisationally closed system is rebuilt by the behaviours it sustains., and vice versa. Consequently, the customary differentiation of structure/behav- iour is arbitrary.

Other !dl!lA Con AT 2.3. Conversation A conversation is sketched in Fig 2. Here, the stable concepts of A and B are "organisa- tionally closed" but also "informationally II (CmlAT) TA open" . Conversation, the act oE concept shar- a description ing, is a pt"ocess of conjoint concept execution, Fi 9 1 a behaviour " by participants A and B, as a result of which the agreed part of the concept is distributed or shared and cannot be func t ionally assigned "to A" or "to B" . A (productive and reproduced) concept (ConAT) in A . !he execution Ex (ConBT)produces-a In Fig 2 participant A is shown as constructing descrlptlon (o r l mage) and possibly a behaviour such as and reproducing a concept for TA (a circl e) driving or drawing circles. Productive and reproductive from concepts for PA (a plane) and QA (a com- operations are also of the type "concept" but act upon pass) ; B constructs and reproduces a concept !&.oAT, TA or other concepts (say where PA i s a plane surface and WAO, where OA- is a compass) to for TB from concepts for RB (cylinder) and reconstruct ConAT SB (slice).

II Mope 01' less independently, I devi3ed P In- These derivations are conveniently dividuation around "[970. Var>ela and Goguen '3 represented by the s horthand notation in Fig "Cl03uI'e" is rruthematical"[y more elegant, Mat - 3. "Urana's autopoiesis is a special case of it, rrunifest in biological Bartlett (55) As a result of agreement both A and B have and Weptheimer ( 56) amongst other'S, are pespon- concepts that are distributive derivations .sibLe for prowNtion and reoroduction. represented by the same shorthand notation , Fig I and Fig 2 can be expanded into in Fig 4, where T*, P*, Q*, R* , S*, are the concurTently opercrting free production systems names of shat""ed concepts. such that the integrity of the production scheme is internally deteI'rllined, and not, ail Under what circumstances may A and B converse usual, determined by an external controZleI', (learn, do each others' intellectual labour, or ppogram. The );Jord means "usu- as in Fig 2)? ally parallel, and 11 ot , as a puLe, confUct free". One scheme for agreement over a siable One prerequisite of conversation 1.S proximity concept is given in "The Organisational Clos- or togetherness. Hut if, as submitted, together- ure of Potent ially Conscious System3" (48) , neSS is increasingly a matter of communication L

A A', Gue,ti"", to B B

E..cutian (Ex) of Execot;"" (Ex) of productive and productive .nd I ""producti ve B' s q.",Uions to A ,," "". ='--. procedure in !&nAT h A's l -explonation r Other " " Other CilllA Coo B ((Q!IAP, !&nAQ) Ow,' ,,",' (CongR. CO"gS) Representative procedure in !&!lgl is S', L-expl.n.tion to A "'"-I b(!&!>ST) _ ' • aescripti"" or'. • description M . • beh.viour • • behaviour TA ,. TB• behaviour or L description - - or L description Fig 2: An A. B. conv"".ti"" in L.ngu.!!'! l •••• result of which, if .gre. .... nt is reached. so,,", of A·, procedure. can eouted .nd .... produced as p.rt of ClwsT and SOIOC of 8's procedure, can be executed and .... produced "' put of CnnA T. 51""01 • • i; i'OIOOrph;'''. TA h port, or .11, of TA, and IS h part, or all. of IS 1()()4 G. paskmlc limits of togethemess

find reasons why A and B will benefit from conversing, as in cooperative action . These reasons are compelling and occasionally sufficient; quite literally A and B must converse if they are to survive. But, just as physical proximity is a specialist answer [0 , the initial question (A and B may converse because they are together), so these constitute very special ised answers to why they must Fig 3: Initial condition of the conversation in 2 converse on some where arcs are derivations and nodes stand for concepts 2.4 Participants

Now, if Fig 2 is taken seriously, t hese questions can be reversed. What are A and B? How are they distinct? Can they exist without distihction? Why should they not share all of their stable concepts, and be as one, on evety occasion? The standard answers, th

The possibilities neither exclude nor derate the traditional demarcation of A and B, but they do open up some real and interesting possibilities such as the following.

(I) A and B are coherent points of view, or perspectives coexisting and interacting in one Shared concept as common to participants brain (for example, Minsky's [57J "proposer" and "critic", apposite in hypothesis formula- tion; my own "teacher" and "learner" appropriate to private study).

(2) A and B are groups of peopl e, teams or societies. (3) A and B are self replicating schools of thought; for example, Lakatos'[5S] "program- mes of scientific research", or the organisa- Fig 4: Result of agreement between tions and cultures evolving in a society. (4) A and B are conglomerates of peop l e and the machinery that exteriorises many of their, normally hidden, mental operations by computing on their behalf. (5) As a speculation, A and B may be collec- tions of interacting but a - priori- independent and computation, then an answer in terms of processors; a computing medium made of neighbourhood (A and H, persons in the same biological or other-than-biological fabric. room) is valid, but exceptionally specialised. Further, if A and B are close for any reason, 2.5 An Observer's Distinction of Participants this does not guarantee conversation. They might, instead, rctain the integrity of An observer may distinguish the participants Leibnizian Monads. It is often possible to • • G. Past/The limits or togethCtneu 1005

• A, B, in any way «I) t o (5) for example) , order to resolve the computati onal conflict pr ovided that the i.solated !,lui tS maintain an proJ:l1bited 1n (al by "esseneiaiN (Nicol1s [ 58 . internal to A o r an internal to B conversation 59] or " cIIscaoded" (PrigogineJ fracess bifurca- (perhaps interpreted as A' 5 thought s and H's tions. (These resemble Thom ' s 60] thoughts). as well as an optional "external" "Catastrophes" but "catastrophe assumes conversation between A and some other unit a canonical state description and says nothing (like 81.As an alternative statement A, B, .. . about the nature of trans! tions between meta- must each be able to adopt more than one stable l'eqionsj .

• perpectio!' (point of view, intention), at once . To pr obe the further , consider one In the role of scientis t s , we ar e anxious to independent processor (one br ain, any inde- observe certain "hard valued" events, like "A pendent part of a brain) able to accommodate a and B agree about T", and to cr ed it them with stable concept; so, for example, the stable factual t ruth (as a luxury, there may also be concept con:espondi.ng to the distributive Fuzzy events based upon the exis tence of a derivation of Fig. 5 certainly makes sense , for "hard valu(!d" substr ate) . This is sur ely Fig. 5 is a copy of the shared part sho",," in possible. Equipment like CASTE includes Fig. 4 with "*M notation removed . The produc- mechani cal dct(!ctor s of evidence for agr eement . tion illl4go

In mechanics. a sentence like "x is a t y", designates a prOposition. In cr. the hard data are sentences " A agrees wi th B over T " (which reflect A and B coherence or agreement ) and designate analogies; very strict ones. For, in order to make the statement at all. the obser- ver must distinguish in any desired way. be- tween A and B . There are indefinitely many ways of dOing so. Relative to this distinction, suppose there is a concept for 11; a construc- but only relative to this distinction, it makes tion in the plane, P, where a many sided sense to assert a similarity of process (in polygon becomes, in the limit, a circle. A the lilll1t, an isomorphism of process); that distributive derivation implicating " coropass some relation proper to A is similar (maybe , and planQ" and "polygon and plane" is shown l.sOQlorphic), to some relation proper to B . This is not a proposition but an analogy in Fig. 6 relation, or, if preferred. a proposition about 8n act of conversation. taken as the basic unit under scrutiny...... The point is fundamental . The episteool ogical framework is changed in order to retain the potential for rigour and for precise stat ement . Within a reflect ive or subjective theory, there is no a priori distinction of s tructure and I. ) behaviour (Section 2.2) but some (any suffi- cient) distinction must be imposed if we are anxious to mako (actually true objective (it referenced) statements.

2.6 Distinction and COnversation 1' 1 Distinctions may be made either by an observer or by the participants - and some must be made if a COherent process is executed, under t he following principles, - 0 - (a) "Coherence depends upon distinct ion".Where- -as many procesaes may , and usually do, give r ise to the same result, one process, executed in the same independent processor, may not give r ise to differont Computationa l -----0'-0 - conflict is disallowed in a unitary (organisationally closed) system.

(b) "Execution of a coherent process in A, B, - 0 - may give rise to distinction (unless ot herwise qualified , to the independence of asynchronous/ Ft95: Instablt in I Mrtni M no tation "nd it. bi f · decoupled parts) ". For example, in a brain, "rtlt'on (d ue to Mt dOro's "Rule of GeM'") T!- T'!- r"and independent/asynchronous regions may ari se as p P'J II'! Tht sJ"IbQ l " 2 I ' lIlI!.ns but the result of t he ongoing activity; in general It (on

Midoro [67] pointed OQt that this violates (a) (t) to (5), or any others of like kind. For if since there is a potential conflict b e tween tnat distinction is not made and if a conver- using a compass and a polygon on the same plane sation takes place. then the bifurcation of P (as an amusing illustration of a fundamental Fig. 6 will necessarily occur T , T" - principle, try to do exactly that. simultaneous- Te" For t he stability criterion or ly) . t10na! closure T T , A B they are t B autonomous (failing a more cogent distinction , Resolution calls for one , or all, of the simply independent) and, at most, isomorphic, constructions in Fig. 6 (a) , 6 (b) , 6 (el, which not identical. introduce independent processors to execute t he relevant concepts, and brains, qua 2.8 SUllmary processors , are able to act in this manner. This is not a postulate about particular Conversation is information transfer between concrete or abstract processes, but a general organisationally closed (alias autonomous) postulate that an asserted process coherence systems. It is a mechanism of conflict requires distinction, her e independence, to resolution, which also generates a distinction support it.. between autonomous individuals to support a • conversa tion. (e) Conversation, either "internal" Or "external" .. is a mechanism of conflict Conversely , if there is no conflict to resolve, resolution ; the act of conflict resolution or there is no need for conversation since there agreement or coherence, is observed in a Petri- are only doppel-gAngers . For example, an ant type information transfer between the partici- does not "see" another ant as an individual. pants, ie. the extent to which the a - priori An obser ver, or an :'\ont- hill organisation D1ay independent asynchronous systems become locally "see" ant:S, as individuals; namely, r obots with dependent/synchronous. common p r ogramming . Something akin to this would be t he fate of mankind if all concepts (d) Alternatively,(Petri t ype) information really had universally agreed definit ions . transfer is conserved and autonomy or dist1nc- tion is generated as a result. An ;!Ippropriate That is one limit of (too much) togetherness, measure Ot transfer 1s Von Foerster's [38] Self an unseemly uniformity . Or ganisation "rate of change of redundancy positive" and "rate of change of energy positi- The other limit of (too much) togetherness , ve" or, strictly (Atkin [62J) a "graded pattern" approached if proximity is enforced by what- over the " simplices" representing the transfer ever means (physical or by communication and relation. computation). is ultra gregarian; a condition that necessarily produces conceptual disparity The informat ion transfer which is conserved in and alienation. In human society it fosters this scheme, is precisely the degree of A'S extremes of hatred or fear. Mankind might, consciousness with a of T (and, vice versa, a's escape its own ravages by intellect ual with A of T); its content being the procedures hermit-hood . Overcrowded rats, are not, exchanged between them. This 1s, after all. the apparently, able to do so. subject matter_ of and as well as "artificial" intelligence although These limits "no need for conversation" and t he mainstream movements in each discipline " no autonomy to allow conversation", parallel carefully avoid any mention of it. If only for the everyday examples of pathology in Section that reason the epistemological trick noted in 1.4. and it as t hough enhanced communi- Section 2 . 5 is neither trivial, nor pedantic. cation (whether by confercncing net works or microprocessors) leads. he1ter 2_7 Automony of individual participants skelter, to one or other of them.

In the concept sharing of Fig. 2 with initial This would be the case if computat ion and conditions shown in Fig. ), and terminal condi- communication parodied in Fig.7 and Fig. 8 , tions shown in Fig. 4, participants A and Bare were simply faster and of greater channel able to distinguish each other in terms of capacit:y . Such an architecture is bound to theil methods of constructing circles. If these produce a large scale version of the miniature stable concepts are, in a limiting case, the difficulties encountered in the cont ext of entire content of A's and S's repertoire, the adaptive training machines and noted in Section labels A and S could be omitted from Fig. 3 and 1. 2 . generated as a result of agreement in Fig. 4 .

Altermltively, of course, the labels A and B may represent all other aspects of A' s or B's conceptual repertoire, in which case, if A can recognize "3", then A sees T* as "a's image of T " and vice verse, if B can recognize "A", ' " tfien B sees T* as "A's image of Ta".

Could A arK!. S have replica methods of making fi9 I: rurlng ""..,i .. circles (or any other idea or theory)? For F S II • flni ... s ...te Port. example, could they both have the circle Storo90 con,1st. in on lofiohe \<1>< .. Ie" co. t>f "", ••d. In.e,lb.d upon oM o..... dtt.n by Tn, derivation of Fig. 5? Yes, if a fSM MjI t>f non oletemini>tic. distinction of autonomy is on different grounds, those of Section 2_3. G. Puk/The limit$ oftogethemess 1007

• • Now, in fact, each vehicle is a lllicroprocessor, parallel- interfaced to the simulated space . Any operation is interpreted and stored in the . . ! . . (no vehicle as a tactic, although tactics may also F S " 1 { :: , be deliberately constructed and called into • execution either by a cOUiDander or by another , tactic, The tactics are programs (for movement, ". action, obtaining information, displaying and storing it) with the usual conditionals, iterations, etc. The commanders do not usually, see tactics as programs, the task is more like navigating or giving instructions of a specific Fig 8, Fra'1"l!nt ot sUlndani coal iOH'Ilcatlon and type. Hence, without being trained programmers, envirOnlilent. the conmanders can due to overload, IDOst) external1se lIlany of their conceptual operations [II< Dec - EnCode r decoder as tactics (alias programs) and the systelll is ...... syntactically band language. conversationally quite powerful . Thus, (al Each commander must maintain a t least two perpectives (at least two vehiCles doing some- -Ther e is nothing wrong with the archit ecture as t hing) a local style; for example, it would be (h) A commander can call for the execution of infuriating if a calculator did not operate as tactics of his own vehicles or of another a degenerate Turing mOdels, perhaps, LOGO programs, written and debugged in two machines, one for A and one Finally, Pig. 9 contains a boX labelled for B. On execution,these programs (the THOUGH'I'S1'ICKER (it is discussed. in the next representative A and B procedures of Fig. 2) section) in which tactics are represented. This give rise to T: and Let the two LOGO system acts in parchanges , • sense, one lIlachine. AS the arrangement. is participants ask for more than this, in parti- purpose built and specialised this claim is not cular an ability to exchange analogic

by Midoro's "Rule of Genoa", inst able concepts). There is neighbourhood within a circumscribed region (each disjoine Circumscription has a distinct. universe of interpretation), but not an absolute hierarchy .

I"terrae. Z 3; , S ., .=, (at In rig J) (u In

, c-nOtr, A "

PermisSible but Quite coorol ex fo rm

..•II ,

(as In 6) prohibited fo .. as result of Rule of SHurcatU Into all fOr.llS shawn transparent to thought, able to accommodat e concepts, and liable to promote convers ation; an information environment of real be ne fit to mankind. , However" if such an information e nvironment is " given (and the arrangements in Fig 9 a nd Fig 12 Fig 10: EnUtlmtnt meshes or conversati ona l dO'lll1 ins do exist) then there 1s a tricky que stion t o answer . Since technol ogy renders communication virtually unlimited , and since a populat i on Personalised hierarchies ari se onl y when pers- architecture renders conversation, without pectives are adopted in order to act, think , or improper discontinuit» quite possible, what learn, as a result of which the me sh is unfol d- exactly does togetherness mean? ed as a "pruning". For e xample , under T , the collection "T, P. Q" (FiglO) becomes T (P.Q), 3 . Conversational Domains under P it. becomes P(Q.T); under T the collect- ion "T, p, Q. R, S" (Figl0) becomes T( (P,Q). (R,S», under P it becomes P(Q,T(R,S» . A speci- What is the domain (envlrorunent) of a conveI"- fic path on a pruning is called a "selective sation? pruning" and is interpretable either as a One answer is participants". specific learning strategy, or a specific Another answer is, "a structure in which plan, or a tactic executed by the vehicle organisationally closed systems of any size are microprocessors in Fig 9, without prior instruct- coherent. unit.s (represented as nodes); the ion and even in the absence of the commander. informat.ion transfer between t.hem is represe n- t.ed by connectivit.y". 3 . 2 'T'HOUGHTSTICKER AND Lp Operations Togetherness, quite fundamentally, means a Entailment meshes are statements in a rudime nta- bet.ween participants, or p e rsona, in ry proto language. Lp, [67, 68] of process cohe- such a connected system. rence and the distinction needed to maint ain i t , 3 . 1 Entailment meshes which delineates the possibilit ies of more re- fin ed expression in a conversational language , Structures of this eype are called "e ntailment L . That. is so, provided there is an op erati on me shes". Some ver y elementary One S are shown i n "saturation" which images, in thi s shorthand . Fig. 10 . Both tho lettered nodes and the notation , the iterative execut ion o f a sta b le concept, or more generally, the ageing and circumscri bed regions are coherent a nd organi - stagnation of a coherent but i s ola t ed o r ganisa- sllt.ionallyclosed (they repr e sen t agreed and tion. "Saturation" (Fig 11) ma.ximises t he deri- stable concepts, or, in the case of b ifurcation Vational redundancy provided that no bifur catiOl , G. Pask{fhe limits of togetherness 1009

> • produced. "Entailment Rings". Bo th of these stable forms > have the important property that adding one more derivation from the same connected region (ie. "Supersaturating" the mesh) produces, for Steiner Systems, a collection of distinct but coherent parts, and for entailment rings, several replicas. p , Contrariwise, the only way to preserve the "1---/ autonomy of a "supersaturated" region is to establish derivations from an otherwise dis- connected region of a mesh which, in this shorthand, stands for a conversation. It in- volves information transfer between organis- ationally closed systems,

F"SM 0

- - -FSI-I n1 F51'! 11 FSM 12 ______J______FSM 12 fSM 22 ------·------fSH n2

, FSM 1m fSM 2m ------fSM om Ini tial Configuration

I

FSM II FSM 21 ------FSM nl

, .. ' , FSM 1m• FSM nm

OpeNtional Configuration Fig 11. Clark's saturation

Fig 12: Organisation of a possibly distributed system for Lp is a k inetic language , not just an alphabet c,,"""rsational interaction that considerably extends the capabilities of Fig 7 or Fig 8. As a useful metaphor the and grammar employed by a user, but a dynamic reserved FSM 0 is introduced at the outset. Sut the fact system, modulated by a user. The systems algo- is that any processor in any (r!1n

4. Evolution of Lp 5. Fundamental Limits

As saturation proceeds, meshes eVOlve!!. Conversation is the stuff of civilised life (asserted in the preface), if only because (in Clark [69] has shown that maximally saturated a precise, and not at all whimsical sense) it (hence, maximally stable organisationally conserves consciousness (Section 2.6 . (d) ) . In closed) forms arc either "Steiner Systems" or the past, limits on conversation were set by too little cOllD11unieation. Nowadays, there is an

ff An implementation of as a proce8S j alao information environment and the relevant limits requires: (a) "Co ndensation" l,)hich cGITies a are imposed by too much togetherness, yielding mesh under the nodes of one 01' severaL per- communication which may appear to be conver- apectives into a node i n a higher-order mesh sation, but not in fact, conversational . (the superimposition of prunings) and "0" node I'; into an analogy. (b) "Expansion" which One meaSure o f togetherness which is apposite retrieves a higher order mesh, creating the only if the transactions are conversational, is universe of interpretation of the analogy "distance" On the metric of a conversational and any disjoint meshes it relates. domain (entailment mesh) which is an expression 1010 G. PaskfThe limitli of togetherness

. or series of expr eSSlons , 1n a protolanguage, (10) G. Pask, The of Human Lp . Learning and Performance, Hu t chi nson, . 1975. The result of saruration and o t her Lp oper- (1i) G. Pask, Conversation, Cognition and ations yields limiting cases. These have in- Learning, Elsevier , Ams t erdam and New terest as being maximally stable if they York , 1975. approach full autonomy in a circumscribed (12) G. Pask , Conversation Theor y : Applications • region (the Lp shorthand for an organisation- in Educarion and Epistemology , ' Elsevier , ally closed dynamic system); beyond that point Amsterdam and New York , 1976 bifurcation takes place unless regions inter - (13) C. Hines, !'a rth act, which is an Lp shorthand image of con- Resources versation. In this sense, conversation is (14) T. Nelson, Compu ter Lib, Hugo Books , "necessary" . Chicago , 1974. (IS) T . Winograd, KRL: A Knowledge Represent- Stared in these terms, which seem a ppropriat e ation Language, Stanford University , 1975. in the context of an informat i on environment, (16) N. Negroponte . Architecture the limits of togerherness are: M. l.T . Press , Bost on, Mass , 1973 , s ee also Soft Archi t ecture Machines, M. I . T . (a) complere saturation (organisational Press and numerous publications , mo s t c l osure and no information tran sfer) ; up to date infOnn.> t ion is containe d in the Quarterly Architecture Ma china tions (b) the type of "supersaturation" t h at yields (17) T . De F3nti, Towards Loopless i nteract ive an indefinitely large number of replicas Programming, proe . Confe rence on Computer (imaging systems which, baing r epl icas, do Graphics Pattern Recognition and Data not need to converse for they have nothing Structure, IEEE Catalogue No 75 , CH0981 . 1C t o say). 1975. (18) R. H. Atkin, Structures in The designers of an information environment Human Affairs, Allen and Unwin , London , would be wise t o avo id these limits , ho... ever 1973. and the limi ts are expressed. Combinarorial Connectivities in Social SyStems , Ileinemann, London , 1977, also REFERENCES Research Reports SSRC HR /021/ 2, Me t hodol- ogy of Q analysis, Dept . Ma t hematics , (1) C. II. Petri, COlYJllunica t ion with Au t omata.. Univer sity of Essex , 1977 . thesis, H.I.T., (trans. F. Greene) . (19) C. Bateson, Steps Towards an Ecology of COR"lllentary in II . Ho lt (1968) Fina l Rept. Mind, Paladin , New York, 1973. for Information System Theory Proj e c t , (20) 1. lllich , IJeschooling Society, Penguin Rome Air Development Centre, Contr act AF 1I0oks , Ilarmondsworth 1973. (for more 30 (602) 4211, 1965. complere reports, Cuernavaca of CIDOC and (2) G. Pask., Adaptive Teaching with adapt- rel a ted publications ; Cuernavaca, Mexico) . ive machines and Electronic Keyboard (21) S . Beer, lIeart of Wiley , teaching machines (reprinted from Jnl . London , Nat . Assoc. Educ. and Commerce), Jul y , 1958 , (22) F. Robb, The Dynamics of Op ini on Change in R. Claser and A Lumsdai ne , e d s . , ,from a system theoretic viewpoi nt , PhD Teaching Machi nes and Programmed Learning Thesis, BruneI University , De p t of Vol . 1. Nat: . Educ . Assoc. Washington, Cybernerics, 1979 . 1960 , 336-366 . (23) M. Robinson , Prisoner's Di lerruna , Meta - (3) C. Pask o The teaching machi ne as a cun- games and other solutions, Behavioural trol mechanism . Trans. Soc . Instr . Tech . Science . 20. 3, 1976 , 201 - 206 (209 as VoL 12 , no 2, June 1960 , .72-89 . reply) (4) G. Pask o An Approach to Cybernetic s , (24) M. Robinson , Angus, Bertha and Conrad : Hutchinson, London, 1961, (rep . '68, 72) . The Cybernetics of Power or Polit ics (5) C. Pask , Adaptive Teaching machines in by any other name, Proceedings, Silver K Aus t wick (ed) , Teaching Machines , anniversary meeting of SGSR, London. Pergamon Press, London, 1964, 79 112 . SGSR , Washington , 1979. (6) B . N. Lewis, G. Pask , The t heor y and (25) S. Braten, The Iluman Dyad , Systems and Practice o f adaptive teaching systems , Simulations, Inst. of Sociology , uni, in R. Glaser (ed) Teachi ng Machines and of Oslo, 1977, also Competing Modes of Learning, Vol. 11, Data a nd Cognition and Conmun icat ion in Simulated Directlons , Nat. Educ . Assoc . Washington, and Self Reflective Systems , Procs. Third 1965, 213-266. ARI Conference on Decision Making 1n (7) C. Pask, B. N. Lewis , The use of a null Complex Systems, Richmond , System Research point method to study the acquisition of Ltd, ( ed. Robinson 1'1. , 1978. and complex transformation skills, (26) V. K. Byshovsky, Control and Informat ion Brit . J. Math and Stat. Psych . 2, 1968, Processing in Asynchr onous Proeessor Ne t - 6l-8lo . works, Proc. Finland USSR Sympos ium on I (8) B. N. Lewis, G. Pask, The self o r ganisation Micro Processors and Data Proce ssors , of a three per son group , in De Brisson Helsinki, Vol I , disc. 1974 . (ed) The Simulation of Human Behaviour (27) C. F. Flores, and T. Winograd, Understand- Dunod , Paris, 1969 , 293-311 . ing cognition as understanding , Stanford (9) G. Pask , B. C.E. Scott , Lea rning and Univeristy Press , 1978. Teaching Strategies in a tran s f ormation (28) II . R. Caines , System Identi f icati on Approx s k ill, Brit. Jnl . Ma t h . and Sta t . Ps ych . imation and Complexity , Man Machine Sys- Vo1.24 . , 1971, 205-229 . tems Laboratory , Universit y of Esse x, 1977 . G. Pask/The limits of togetherness 1011 •

(29) : Foundation and eds, How Stude,nts Learn, Lancaster I nst. .. Practice, 3rd ARI Richmond Educ.. Landlster. Making in Complex (46) N. J . Entwistle, The Verb to Learn takes Systems, M. Robinson, cd., System Resear- the accusative, Symposium, Learning Proc- ch Ltd . 1978 . esses and Strategies, Br. J . Educ. Psychol (30) T. Gergely and L. Ury, Nondeterministic 46, 1976 , 1-3 and Knowledge Structures Programming within the frame of firs t and Strategies, Sr. J. Educ. Psychol, 48, order classical logic, Parts I and II, 1978 , 1-10. Hungarian Academy of Sciences, Budapest, (47) G. Pask, Conversational Techniques the J 9 79. Study and Practice of Education , Br. J. (31) T. Gergely, and}L Szots, On t he incom- Educ. Psychol, 46, 1976, 12-25. and pleteness of prov.Lng partial correctness, Styles and Strategies of Learning, Br. Acta Cybernetica, Tom, 4 , Fasc. 1, J . f:duc. Psycho, 46, 1976, 128-148. Szed, 1976, 45-67 . (48) G. Pask, A Conversation Theoretic Approach (32) R . Glanville, What .LS memory that can to Social Systems, F . Geyer and J van der remember what it is?, Proceedings, 3rd Zouwen eds, Soci o Cybernetics: an actor EMCSR, Vienna, rep . Vol IV Progress in oriented social , Martinus Cybernetics and Systems Research, R. Nijhoff, Amsterdam, 1979, 15 26. Trapp! and G. Pask, eds. Halstead Press, (49) G. Pask, et al Developments Conversat- New YOTk, 1976 , 27-37. ion Theory: generalisations and language "33) R. Clanville, The Concept of an Object structures, Part I to VIII (parts I to V and the Object of a Concept, Proceedings completed),. In. J. Man Machine Studies, 4th EMCSR, Linz; Hemisphere, Washington, press . 1978, press. (50) G. Pask, Organisational Closure of Potent- 134) J. A. Goguen, J . l,r. Thatcher, E. G. ially Conscious Systems, NATO Congress on Wagner, and J. B. Wright, Initial Algeb- Applied General Systems Research, ra Semantics, Mathematical Sciences hamton, 1977, in Zeleny ed, Dept. IBM, Thomas J, Watson Research North Holland, New y,;, Center, Yorktown Heights, New York, 1976 . press. (35) F. Varela, r:rinciples of Biological Auto- (51) J. Von Neumann , Theory of Self Reproducing nomy, Elsevier, New York, 1980 . (this Au tomata, edited and completed by A. Burks, book contains detailed source references University of Illinois Press, Urbana, 1966. to Varela and others). (52) A. W. Burkes, ed, and contributor, Essays (36) H. R. Neurophys iology of Cognit- on Ce'1lular Automata, University of Illin- ion, Cognition, A Mul t iple View, p, L. Press, Carvin ed, Spartan Books, New York, 1969, (53) L. Lofgren, An Axiomatic Explanation of rep. Cybernetics of Cybernetics, BeL, Complete Self reproduction, Bull, Math. Urbana, 1974. Biophysics, 30, 1968, 415-424 (37) W. S. McCulloch, Embodiments of Mind, (54) L. Lofgren, Relative Explanations of MIT Press, Boston, Mass, 1965 Systems, in G Klir ed, Trends in General (38) H. Von Foerster, Molecular Ethology , Systems Theory New York , 1972. C. Ungar ed, Molecular Mechanisms of (55) F. C. Bartlett. Cambridge Memory and Learning, Plenum Press, New University 1932. York, 1971. (56) M. l,lertheimer, (39) H. Von l'oerster, An Epistemology for Social Science , Living Things, E. Morin, ed, L'Unite de (first pub . 1945). l'Honnne, Editions de Sevel, Paris, 1976. ( 57) M. Minsky, A Framework for representing (40) H. Von Foerster, A. I nselberg, P, Weston, knowledge , in P. Winston, ed., The Psycho 1 Memory and Inductive I nference, Cyhernet- ogy of computer Vision, McGraw Hill, New ic Problems in Bionics, H. L. Oestereicher York, 1975. < and D. R. Moore, eds., Gordon and Breach (58) I. Lakatos, The Methodology of Scientific London, 1968. Research Programmes, Vol 1. T. Worall and (41) H. Von What is memory that it G Currie, eds., Cambridge Univeristy , and foresight as well? Press, London, 1978 (collected papers). S. (59) J. S. Nicolis, E. N. Protonorios, and , 1969. Theologou, Comrm{nication between two self (42) 1I. Von Foerster, On Sel.[ organising Syst - systems modelled by controlled ems and their Environments, M. C. YQvitts Markov chains, Int. J. Han Mach. Studies, and S. Cameron cds, Self Organising Sys- 10, 4 , 1978, 343-367. terns, Pergamon, New York, 1960 . (60) G. Nicolis and I Prigogine, Self Organis- (43) H. Von Foerster, and G. Pask, A Predictive ation in non equilibrium systems, Wiley, Model for Self Organising Systems, London, 1977 . no 4, 1960, 258-300, and (6t) R. Thorn , Structural Stab ility and Morpho , , 26-55. .g.enes.Ls, Trans, D. H. Fowler, W A Benjamin (44) G. Pask , D. Kallikourdis, B . C. E. Scott, Inc . 1975 . A Fresh Look at Cognition and t he Indiv- (62) R. Atkin, A Kinematics of Decision· Making idual, (initial paper of 5) Int. J . Man in Procs. 4th ARI Richmond Conference On Machine Studies, Vol, 4, 1972, 211- 2 16; Decision Making in Complex Systems, Vo l. 4, 1972, 217-253; Vol 5, 1973, 17-52, M. Robinson and G Pask, eds, ARI Tech. Rep. Vol. 5, 1973, 443-566; Vol 17, 1975, 1979, (see also, R. Atkin, Time as a 15-134. pattern on a multi dimensional structure, (45) J. Daniel, Conversations, individuals J. Soc. BioI. Struct. 1, 1978 , 281-295). and knowables, Eng. Educ. Feb, 1975, 415- (63) S. Papert, Teaching Children Thinking, 420, also N. J. Entwistle, Hounsell, D, Procs, IFU Conf. Computer Education, 1012 G. Pask{The limits or togetherness •

lunsterdam, 1970 . (Reprinted in Mathemat- ics Thinking. Bull. Assoc. of Teachers in Mathematics , No. 58, 1972) . (64) C. Pusk, Observable Components of tbe decision process, and a revised t heoret- ical pOlition, in Proceedings, 3rd ARt Richmond Confer ence on Decision tn Complex Systems , M. Robinson, ed., ARt Technical Report , Washington. (65) S. B. Hansen, The Architecture of Con- • current Programs , Prentice Hall, London. 1979. (66) M. MinSky , J. Howe, Personal Communication and disoussion of draft papers , 1979, 1980 (67) C. Pask , Kinetics of Language illustrated by a Protologic Lp, Proceedings, Workshop on Formal Fuzzy and Cognitive Processes, 2nd Congress Int. Assoc. for Semiotic Studies , Vienna, reprinted in Semiotics, Benjamins, Amsterdam, 1979, (in press) . (68) C. Pasko An Essay on the kinetics of lang- uage • behaviour and thought, Proceedings Anniversary Int. Meeting, Society for General Systems Research, London. SCSR Washington , 1979 , 111-128. Alia in same volume, Against Conferences , Poverty of Reductionism. (69) P. T. Clark, Saturation, Bifurcation, and Analogy in Lp , Proceedings , 4th ARt Richmond Conference on Decis i on Making in Complex Systems , C. Pask ed . , ARI Technical Report , Washington, 1979, (in pren) .

•