The Act of Creation

by Arthur Koestler

  Thus all along the line we abstract and discriminate only qualities which are relevant to us; and new discriminations arise as a result of changes in our criteria of relevance -- where 'us' refers to animal and man. In the normal development of the individual these changes are due to maturation and guided learning. In the experimental laboratory, as in reformatory schools and other brainwashing establishments, rewards and punishments effect the transformation of the subject's, or victim's, scale of values. In classical conditioning all tuning forks are 'the same thing' to the dog until the difference in pitch is made significant by the giving or withholding of food. In guided learning, all pages in his algebra book look the same to the schoolboy until he learns to distinguish linear from quadratic equations at a glance.

  Distinctions which are irrelevant to the subject will either go entirely unnoticed; or they may be perceived but not retained; or they may be retained but not verbally discriminated. Thus if we fail to observe differences, it is either because we lack the equipment for doing so ('lack of discrimination' in the colloquial sense), or because the differences are 'indifferent' to us, or not important enough to give them verbal labels (e.g. knowing stars by sight but not by name).

  I have dwelt at some length on the subject of abstraction and discrimination because although most of what I have said would seem self-evident to the layman, this would not be the case with students of experimental psychology -- or of philosophy, if it comes to that. Already in the twelfth century A.D. John of Salisbury remarked that the world had grown old discussing the problem, and that it had spent more time on it than the Caesars took to conquer the world. But Pavlov, Watson, and those directly or indirectly influenced by them have certainly made confusion worse confounded.*

  From the genetic point of view, abstraction and discrimination appear (cf. pp. 608 f.) as the latest extensions of the basic principles which we saw at work on all levels of the hierarchy, starting with the integration of functions and differentiation of structures in morphogenesis. Abstraction, by creating pattern and order out of the chaotic stream of experience, corresponds to the former; discrimination in perception and the consequent differentiation of response correspond, as language indicates, to the latter. If we wish to indulge in analogies, we might say that the categories of Aristotle acted as embryonic inductors on the self-differentiating morphogenetic fields of conceptual thought.

  The Magic of Names

  We have so far distinguished several stages in the child's progress: (a) the abstraction of pre-verbal object concepts (and action concepts); (b) attaching a verbal symbol, which soon acquires central importance, to the concept; (c) the discovery, signalled by the appearance of the naming question, that all things have names, that the words previously learned are only particular instances of a general relation between words and things. The next step consists in the concretization of this relation itself in the concept 'name'. The child has realized that not only have all things verbal handles and labels attached to them, but that these labels and handles are called names.

  A characteristic feature of this development is that at each step a relation has been abstracted and turned into a relatum. At the first step the relation between the varied particular appearances of the mother was turned into a single relatum 'mama', which now enters as a unit into other relations; at the last step the relation between words and things was abstracted and turned into the conceptual entity 'name'. 'Name' is the verbal symbol attached to the relation of verbal symbolism; by being made explicit and conscious the relation is now experienced as a concrete relatum.

  In fact, an over-concrete relatum. To quote Piaget: 'Names are, to begin with, situated in objects. They form part of things in the same way as do colour and form. Things have always had their names. It has always been sufficient to look at things to know their names. . . . To deform the name is to deform the thing.' [14] When a child of four and a half was asked how one knows that the sun's name is 'sun', it answered: 'Just because one sees it'. And when a child of nine was asked whether one could have given another name to the sun, he answered: 'No -- because the sun is just the sun'. Another child of six and a half, when pressed, admitted that God could have given the sun another name, but in this case 'God would have done something wrong'. [16]

  When Herschel discovered Uranus, the German naturalist Sachs remarked sceptically: 'What guarantee have we that the planet found by him really is Uranus?' Equally inspired was this philosophic reflection of an Englishman: 'English is the most logical language; a knife, for instance, is called by the French couteau, by the Germans Messer, and so on, whereas the English call it a knife which is after all what a knife really is.'

  In the mentality of primitives, the person and his name are magically related. In Eastern religions, evocation of the names of deities -- the recital of mantras -- fulfils a magic function; in Tibetan Buddhism the work is left to the prayer mill. This attitude lingers on in medieval philosophy (Realists versus Nominalists); in all forms of magic, and, more covertly in modern science -- in the unconscious belief that words like gravity, entelechy, or electro-magnetic 'field', etc., somehow have an explanatory value an sich (cf. Book One, VII). Such is the power of verbal symbols to focus attention that it confers on hazy concepts in statu nascendi the appearance of hard, tangible concreteness, and 'gives to airy nothing / a local habitation and a name'. The name is then experienced as a self-evident explanation, a saturation of free valencies as it were.

  The Rise of Causality

  During its first years the child does not discriminate between nominal, attributive, and causal predications -- as earlier it did not differentiate words according to grammatical categories.

  When children between five and six are asked: 'Why does the sun not fall down?', they will answer: 'Because it is hot', 'Because the sun stops there', 'Because it is yellow'. [15] And the moon does not fall down 'Because it is very high up', 'Because the sun is not there', etc. The significant aspects of an experience are connected as 'going together' in an undifferentiated 'feeling of relation'. [16] Goethe's 'Connect, always connect' seems to be the motto of the child as, out of the fluid raw material of its experiences, it selects and shapes patterns and relations -- relations which will be re-classed and re-grouped later on according to shifts in motivation and interest leading to the emergence of new criteria of relevance -- until the final, more rigid but not always more perfect adult relation-categories emerge. The urge to connect, to aggregate matrices of experience into more comprehensive ones; the fumbling for hypotheses about the way things are held together, the tentative formulation of rules of the game -- in all these fertile activities we see the participatory tendencies at work: intimations of the fundamental unity of all things. Later on they will crystallize in magic causality, with its correlates: animism and 'mystic participation' (to use an expression coined by Lévy-Bruhl for the mentality of primitives, and applied by Piaget to the mentality of the child). Needless to say, the self-asserting tendencies too play their obvious part both in the child's overt behaviour and its fantasy world.

  It seems that the first relational patterns which are discriminated are relatively static forms of attribution (of names and other properties), and of dynamic changes-in-time. The latter give rise to a vague 'feeling of causal relations' [20] derived from the cumulative experience that 'things make other things do things'. At this stage, word-classes begin to emerge which roughly correspond to substantive-nouns, adjective-attributes, and action-words or verbs. But these classes, and the types of relations implied in them, remain for a long time fluid. The child's progress towards grammatically more correct forms of speech is mainly due to imitation and conventional training -- which mask the fact that behind the increasingly adult forms of expression, magic ways of thought survive. They survive, of course, even in the adult, and never vanish completely. Thus the stabilization of the codes of grammar and syntax in no way corresponds to the dynamic evolution of thought, and inferences drawn from the former to the latter have fo
r a long time misled child psychologists. [21]

  With the momentous realization that 'one thing leads to another', intimations of causality emerge from the fluid pool of perceived 'togetherness' and 'relatedness'. The "homo novus" has now set out on a long and tortuous road, which has in fact no end, except where, of his own choice, he will come to rest -- theological causality, mechanical causality, statistical probability -- gods playing billiards or throwing dice. But children are philosophical optimists, and the same process of empirical induction which earlier on led to the discovery that all things have names, now leads to the discovery that all things have 'becauses'. The sequel is a familiar one: the naming question is replaced by the 'why' question which is many a parent's despair; just as a thing was incomplete without a name, so now an event has a free valency which must be filled by a 'because'. The actual content of the proffered explanation, the when, and how, and how much, is as yet of secondary importance; 'the sun does not fall down because it is yellow' is accepted as quite as satisfactory an answer as 'the dolly is called dolly because that's her name'. Just as, at the earlier stage, when a name was not immediately offered, it was invented to fill the vacancy, so now, when no 'because' is proffered at once, an expiation is provided ad hoc by the child itself. And just as the invention of names is guided by onomatopoeia or private associations, so the causal explanations of children are derived from their private matrices of thought.

  The questioning mania is reinforced by the desire to attract attention and to be in the centre of events; it is a typical mixture of participatory and self-assertive motivations. Lorimer made a record of all the why's asked by a child of four years, eleven months, in the course of four days [17], which is both charming and instructive. (One asterisk signifies that no answer was given; two asterisks: no answer given, question not repeated):

  1. Why do you have this box for your feet? 2. Why did they bring the bed down from the attic? 3. 'Take your dollies in now, Joyce!' Why? 4. 'Fix the rug! You caught your feet under the edge of it!' Why? Why did I? 5. Why did you take two cookies? 6. Why does the watering-pot have two handles? 7. Why did he put the solder in so many places? 8. 'The song-sparrow isn't pretty to look at' -- Why isn't he pretty? 9. 'The bobolink has a brown coat in winter.' Why?* Why? 10. 'It was careless of you to lose your shovel.' Why?* Why? 11. 'This is your orange juice.' Why?** 12. 'You are to sit here, in Daddy's place.' Why?* -- Please, Mother, tell me why. 13. Do we have bangs at the back of our heads? 'No!' Why?* Why, Mother? 14. 'And then he made a mast for his little boat.' Why? Please tell me why he made a mast. 15. Why do you wash the hair off (the razor)? 16. (Putting on bathrobe without putting her arms in the sleeves) Is this good? 'No.' Why?* -- (impatiently) Why? Speak out! (but then without waiting for an answer) Because it wouldn't stay on. 'There, you thought it out for yourself, didn't you?' 17. Why did you stub your toe? 'Because I wasn't watching out.' Why? 18. 'I will tell you a story about this willow plate' -- Why?** 19. 'Please hurry, Joyce!' Why?* -- Because you want me to wash?* Why didn't you wash first? Because you knew I wanted to go with Daddy? 20. See the little tea things! Why did we buy them? 'Why do you think?' Because we might use the others all up. 21. Why did you use both a fork and a spoon in making that cake? 22. Why did you sit in that chair, Mother? 23. 'Please don't climb in that chair!' Why?** 24. Why are you putting up that screen?** 25. Why are you opening that window?** 26. Why does the little chicken grow in the shell? 27. 'You can't win by jumping up and down!' Why? 28. Jeremiah, Jeremiah. He got into a pit, didn't he? Why did they put him into a pit? 29. 'Please be careful not to break the bean-plants.' Why? 30. I saw your blue apron through a crack in the door. I thought it was a spider. 'A spider isn't blue, dear!' Why?* -- Please, Mother, tell me why a spider isn't blue. 31. 'You shouldn't talk about a visitor's beard, Joyce, until he has gone!' Why?* Please tell me why. 32. Why don't you have a beard, Mother? 33. I want to cut my eyebrows in half! 'Oh! You wouldn't want to do that!' Why? Because I would look funny? 34. Why do we have eyebrows? 35. Why must I hurry? 36. Why should I wait for candy until after supper? 37. Why did you speak to that man? 38. 'Please don't bang the car-door!' Why?* 39. Why did the chickens walk in front of the car? 40. 'It is time to go home for dinner now!' Why?*

  A certain number of these questions are obviously motivated by the desire to attract attention or intended as a protest; others are quasi-automatic exercises of the questioning habit -- they remind one of Leerlauf activities in vacuo. But others, such as Nos. 2, 9, 26, 32, are expressions of genuine curiosity; Lorimer put thirteen out of forty questions into this category, judging them by content and the child's expression. It is curiosity of a new type, no longer directed at the practical or playful uses of things only, but at the mystery of their 'becauses'.

  The word 'because' now plays a similar part to that which the word 'name' did before: an abstracted relation has become a relatum, concretized in a verbal symbol. The child's concept of 'becauseness', i.e. causality, will undergo a series of changes, but not the verbal symbol which refers to it. Later on, the causal relation will enter as a relatum into the higher matrix of 'logical categories'; and even later this class, in its turn, will become a member of the matrices of epistomology, psychology, and so on.

  Explaining and Understanding

  This leads us to the question of the nature of explanation.

  Earlier on I quoted Craik's suggestion that the nervous system's main function is 'to model or parallel external events', and that 'this process of paralleling is the basic feature of thought and explanation'. [18] In terms of the present theory the 'model' consists of hierarchies of flexible matrices with fixed codes, abstracted by the organism according to its lights. Insight and understanding then become relative terms, the degree of understanding depending on how many different aspects of reality have been abstracted, how sharply they are discriminated, to what extent the abstract codes lend themselves to explicit formulations, and the degree of precision and error which the model reveals when subjected to the test of empirical verification.

  We have seen that it is necessary to distinguish between progress in understanding -- the acquisition of new insights, and the exercise of understanding at any given stage of development. Progress in understanding is achieved by the formulation of new codes through the modification and integration of existing codes by methods already discussed: empirical induction, abstraction and discrimination, bisociation. The exercise or application of understanding -- the explanation of particular events -- then becomes an act of subsuming the particular event under the codes formed by past experience. To say that we have understood a phenomenon means that we have recognized one or more of its relevant relational features as particular instances of more general or familiar relations, which have been previously abstracted and encoded.

  The conventional test of understanding is verbal explanation -- the subject is invited to name the general rule of which the event to be explained is a particular instance. But the availability of such neat and ready explanations is the exception rather than the rule -- unless the explanation was learned by rote -- because, in the first place, the codes which govern perception and cognition function below the level of focal awareness; in the second place because a great number of codes which govern thinking are unverbalized -- including the codes of verbal thinking, grammar, and syntax; thirdly because there are emergent, 'nascent' codes which are still unstable and cannot be 'pinned down', but are sometimes nevertheless of decisive help to understanding. We thus arrive at a whole series of gradations in understanding and explanation -- such as:

  (a) Unconscious understanding mediated by the dream -- a form of internal discourse in which specific experiences are subsumed under very old, emotion-charged matrices with pre-verbal codes. The transformations and disguises which people and events undergo in the dream may be described as acts of recognition of different appearances as the 'same thing' on the scales of symbolic relevance peculiar to the dream. Myth, folklore, fairy-tale, the fantasy world and magic caus
ality of the child are mainly inspired by this type of understanding; and the explanations offered by primitives and children for their beliefs are true explanations in the sense defined.

  (b) Tentative explanations, which indicate that the matrix into which the event is to be incorporated is still in the process of construction by trial-and-error learning and hypothesis-formation.

  (c) Half-understood explanations referring to matrices in "statu nascendi" which, unlike (b), are being formed mainly by unconscious guidance, by unverbalized analogies, etc.