The Denial of Creavitity
Nearly half a century has passed since the publication of Watson's Psychology from the Standpoint of a Behaviourist (1919), and few students today remember its contents or even its basic postulates. In Watson's second book (1924) there is a chapter entitled: 'Talking and Thinking -- Which, When Rightly Understood, Go Far in Breaking Down the Fiction That There Is Any Such Thing As "Mental" Life.' In this chapter, the behaviouristic view on the creative activities of man is set down in a simple and striking way (all italics are Watson's):
How the new comes into being: One natural question often raised is: How do we ever get new verbal creations such as a poem or a brilliant essay? The answer is that we get them by manipulating words, shifting them about until a new pattern is hit upon. . . . It will help us to go to manual behaviour. How do you suppose Patou builds a new gown? Has he any 'picture in his mind' of what the gown is to look like when it is finished? He has not. . . . He calls his model in, picks up a new piece of silk, throws it around her; he pulls it in here, he pulls it out there, makes it tight or loose at the waist, high or low, he makes the skirt short or long. He manipulates the material until it takes on the semblance of a dress. . . . The painter plies his trade in the same way, nor can the poet boast of any other method. [6]
The key word is 'manipulation', defined by Watson as an 'instinctive tendency sometimes exalted by calling it constructiveness. That there is an original tendency to reach out for objects, to scrape them along the floor, to pick them up, put them into the mouth, to throw them upon the floor, to move back and forth any parts which can be moved, is one of the best grounded and best observed of the instincts.' [7] That is all we learn about manipulation. It is a random activity, which, through elimination of useless movements by the trial-and-error method, gradually develops into ordered habits. The rat, put into an unknown maze, goes on manipulating its motor-organs at random until it hits upon the food in the same way as (the comparison is Watson's own [8]). Patou manipulates the piece of silk until he hits upon a new model; likewise, the poet or essayist 'shifts about' words 'until a new pattern is hit upon'. It is expressly denied that Patou, the poet, or painter has any 'picture in his mind' of the kind of thing he is planning; he simply goes on manipulating his units until the model, poem, or drama is 'hit upon'. At that moment the stimulus 'to arouse admiration and condemnation' ceases to be active, and manipulation stops -- 'the equivalent of the rat's finding food'. [9]
Obviously, viewed from this angle, psychology presents no problems. Yet the matter is of more than historical interest, because, although the cruder absurdities of Watsonian behaviourism are forgotten, it had laid the foundations on which the later, more refined behaviouristic systems were built; the dominant trend in American and Russian psychology in the generation that followed had a distinctly Pavlov-Watsonism flavour. 'Each of these systems', Hilgard wrote (referring to the immensely influential schools of Guthrie, Hull, and Skinner*) 'represents in its own way a fulfilment of the behaviouristic programme originally proposed by Watson.' [10] The methods became more sophisticated, but the philosophy behind them remained the same. Originality and creativeness have no place in it.
For Guthrie, the original solution of a problem 'must be in the category of luck, and hence lie outside of science'. [11] In Skinner's works, as I have just said, the word 'insight' does not occur; and the technique of problem-solving is, in Skinner's view as in Watson's, 'merely that of manipulating variables which may lead to the emission of the response. No new factor of originality is involved.' [12] Hull expressly postulated that the differences in the learning processes of man and rat are of a merely quantitative, not of a qualitative order:
The natural-science theory of behaviour being developed by the present author and his associates assumes that all behaviour of the individuals of a given species and that of all species of mammals, including man, occurs according to the same set of primary laws.' [13] The verbal and mathematical symbolism of man, verbal communication and written records, were considered to differ only in degree, not in kind, from the learning achievements of the lower animals, epitomized in the bar-pressing activities of the rat. 'Hull did not intend merely to systematize the account of rat lever-pressing, from which most of the data for his later set of postulates derived. He intended to arrive at the basic laws of behaviour, at least the laws of the behaviour of mammalian organisms, including the social behaviour of man. [14] **
The state of affairs in that period has been succinctly summed up by Osgood:
Lloyd Morgan's canon -- that the behaviour of animals should not be explained in terms of human attributes if it can be explained on a lower level -- was designed to counteract the common tendency to put oneself in an animal's place and explain its actions in terms of what we would do in that situation. . . . It is interesting that through the behaviouristic phase in which American psychology has been moving, Lloyd Morgan's canon has been subtly inverted. Many present-day psychologists are loath to attribute to humans any characteristics that cannot be demonstrated in lower animals. [15]
In other words, for the anthropomorphic view of the rat, American psychology substituted a rattomorphic view of man.
The Advent of Gestalt
A turning point seemed to have been reached with the publication of Köhler's Mentality of Apes in 1925. The Gestalt school had been steadily growing in Germany since Wertheimer's first papers in 1912. As already mentioned, the German edition of Köhler's ape book appeared in 1917, but the first American translation only eight years later. Imagine Einstein's General Relativity theory, also published in 1917, reaching America with an eight years' delay! Yet physicists are supposed to have a limited, psychologists a broad and open-minded, outlook.
The historical merit of the Gestalt school was, beside its concrete discoveries about perceptual organization, to crystallize the convergent trends towards a new, dynamic conception of the organism as a living whole -- and not merely as the sum of its parts. Such trends had been developing since the beginning of the century, independently from one another, in biology, embryology, neuro-physiology, and in psychology itself. But only when the new terms 'Gestalt', 'configuration', and 'functional whole' became fashionable slogans comparable to Relativity and the Oedipus Complex -- did this silent revolution penetrate into the broader public's mind and convey some vague idea of a new orientation towards the problems of organic life and the human intellect.
However, the great expectations which Gestalt aroused were only partly fulfilled; and its limitations soon became apparent. Gestalt-explanations seemed to flourish only in the area of their origin, visual perception; when it was attempted to transplant them into the fields of cognition, memory, neuro-physiology, even the other sense-modalities, they seemed to wilt away. Some of these limitations I have already mentioned in previous chapters; the ambiguities of the central concept of 'insight' will be discussed in the next one. The result was a kind of abortive Renaissance, followed by a behaviourist Counter-Reformation. The neo-Behaviourists, having incorporated some of the Gestalt findings into their theories, had indeed a remarkable come-back; the Gestaltists remained more or less firmly entrenched in their positions which displayed 'good closure' all round. The experimental evidence was mostly inconclusive; some of it proved damaging to one school, but without directly confirming the contentions of the other. Each camp was divided in itself; and (apart from the 'lone voices' of the elder generation) a 'third force' began to make itself increasingly felt, comprising such outstanding freelances as Tolman and Hebb, who stood with one foot in each camp, as it were. Hebb has compared the situation to 'the running battle between the Left and Right' [16] where each party, while shouting its own slogam, tacitly keeps adopting ideas originally advanced by its opponents. All this, of course, refers to the American scene; but in England, at least, developments followed similar lines.
In spite of this rapprochement, and the new outlook of a younger generation (cf. the chapter on Motivation), some basic
differences still divide learning theory into two broadly outlined camps -- differences not on points of fact, but on their interpretations -- on explicitly stated or tacitly implied axioms, general outlook and selective emphasis. These can be briefly schematized as follows:
S.-R. Theories contra Cognitive Theories ______________ __________________ Conditioning Insight Chained responses, Patterned, flexible responses stamped in bit by bit adapted to the total situation Gradual learning by Sudden learning and trial and error problem-solving through insight. Acquisition of habits and Acquisition of knowledge skills through reinforcement ('cognitive structures') through latent learning Emphasis on peripheral, Emphasis on central sensory motor activity cognitive processes Emphasis on discrete Emphasis on relation-patterns, stimuli, on parts and wholes, perceptual Gestalten perceptual elements Motivation = reinforcement Motivation by exploratory by need- or drive-reduction, drive, or its combination or anticipation thereof with other primary drives Continuous linear gradient Hierarchic levels of organization leading from rat to man
This schema follows (except for the last two points) by and large Hilgard's classification of 'issues on which learning theories divide'. [17] Only a few prominent psychologists would subscribe to all the principles listed in either of the two columns; but a majority of them would probably subscribe to the majority of the principles in a single column.
The core of the controversy could be summed up in shorthand as 'drill' versus 'insight'. The answer, already suggested, seems to be that the various methods of learning form a continuum extending from classical conditioning at one end to spontaneous, intelligent problem-solving at the other, while in the intermediary ranges we find various combinations between drilled-in and insightful learning, depending on the animal's ripeness for the task to be learned. This approach, which aims at synthesis, not compromise, is of course by no means original; it is shared -- though for somewhat different reasons -- by 'functionalists' like Woodworth, behaviourists like Hebb, and ethologists like Thorpe. Thus Hilgard wrote on the 'functionalist outlook': 'Learning is not blind on the one hand and insightful on the other; there are degrees of understanding involved from a minimum at one extreme to a maximum at the other, with most cases falling between these extremes.' [18] However, the definitions of 'insight' and 'understanding' vary, which leaves us with the same problem, only in a different formulation. Let us try to get closer to it by considering some typical examples of animal learning.
Conditioning and Empirical Induction
A newly hatched chick will peck at grains, worms, caterpillars in its neighbourhood. If a so-called cinnabar caterpillar is now placed before the chick, of which species it has no previous experience, it will peck at it as at any other small object, but reject it at once with signs of distaste. With the majority of chicks one single experience is enough to make it in the future avoid caterpillars by sight. (The cinnabar caterpillar has a distinct black and gold colouring, a visual pattern easy to retain.) Thus the chick has acquired a new skill, the avoidance of caterpillars, after a single experience (or one repetition in the case of less gifted chicks); and moreover, that skill, or 'cognitive structure', or knowledge, or whatever you call it, is a correct replica, in the chick's nervous system, of the relation between the visual appearance and disgusting nature of caterpillars. Must we ascribe insight to the chick; or shall we adopt the opposite viewpoint, according to which the single try was sufficient to eliminate the error of pecking at caterpillars and to establish the avoidance reaction; and are we faced with a real alternative or merely a verbal quibble? Let us, for the time being, leave the question open.
A dog is an animal of much greater inteljigence than a chick, and yet in Pavlov's laboratory dogs require long series of repeated experiences for learning to relate certain perceptual signals to the imminence of food. Weeks of stamping-in are often necessary to make the dog differentiate between the food-signal values of a circle and an oval, whereas a single experience is sufficient to make the chick differentiate between the signal values in the appearances of a worm and a caterpillar.
The reason for this contrast has already been discussed (Chapter X). Caterpillars belong to the natural environment of the chick; the perception of the striped creature is a biologically relevant input while the chick is engaged in pecking; its horrid taste makes it even more relevant; the visual input will accordingly be allowed to pass through the filters of the memory hierarchy, where it will be encoded and serve as an analyser-device for future inputs. On the other hand, gongs, bells, metronomes, tuning forks, cardboard figures, red lights, and electric shocks have no biological relevance to the species dog, nor to the individual dog outside the laboratory. In its natural environment the dog would pay no attention to them, but pursue some exciting scent; the cardboard ellipse would never have a chance to form a stable trace in the dog's perceptual organization.
How should one explain, then, that the experimenter nevertheless succeeds in stamping in the response? In the first place, a Pavlov dog in its restraining harness is not a dog, but a preparation, which can only be found "in laboratorio". It is immobilized on the experimental platform, in a soundproof laboratory, alone, cut off from all natural stimuli and habitual activities; it is, so to speak, isolated under a glass bell. This state creates a particular stress in the animal, called the dog's 'laboratory attitude', which is sharply distinguished from its behaviour outside the lab. Deprived of all other stimuli and activities, the ticking of the metronome or the figure on the cardboard are the only events on which the dog's attention can focus; there is no competition between different inputs; and thus the originally irrelevant stimuli are gradually transformed into relevant stimuli and encoded in stable traces. Relevant to what? To the only biologically important events which are allowed to occur under the glass bell: the periodic appearance of meat-powder by remote control. The dog's laboratory attitude is dominated by this periodically repeated event; and as his perceptual hierarchy becomes slowly readapted to pay attention, in the absence of other stimuli, to the irrelevant sound of the bell, the nascent trace of the bell-sound will become incorporated into the feeding hierarchy. If the sound of the bell always overlaps with the appearance of the meat-powder, then the sound will eventually trigger off the feeding code, as the first two bars of the Marseillaise will trigger off the following bars; the dog will begin to salivate. But salivation is merely the first, anticipatory act of its feeding behaviour, and if no meat-powder is actually presented, it will stop there; the dog will not chew and snap at nothing. In the absence of food, the feeding habit gets no 'environmental feedback', and the activity comes to a halt at the expectant, salivatory stage. It is quite untrue, therefore, to say that the 'conditioned stimulus', e.g. the bell, has been 'substituted' for the 'unconditioned stimulus', the meat-powder. What happened was that the dog has learned, by the cumulative effect of its past experience, to expect the appearance of the meat-powder after the bell, because that is the 'rule of the game'. He salivates, not because he confuses the bell with food, but because he expects the food, signalled by the bell. We can say, with Polànyi [19], that the dog has arrived at a correct empirical induction; or with Craik, that the dog's nervous system is now functioning 'as a calculating machine capable of modelling or parallelling external events' which is 'the basic feature of thought and of explanation' [20]; or in our own terms, that the invariant factor in a repeated sequence or experiences has been encoded in the dog's brain.
All this is a far cry from the conception of reflex arcs in which USs, CSs, URs, and CRs are mechanically coupled together or substituted for each other like railway carriages in a shunting yard. In fact, the dog's behaviour in the strange, artificial universe where red lights portend food and metronomes electric shocks, is eminently 'logical'; and the reason why it takes so long to stamp in the lesson is that the dog must readapt its entire attitude and hierarchy of values -- of what is important in life and what is not -- to that universe, where natural law is replaced by Pavlov's law; a kind of Ni
etzschean "Umwertung aller Werte". Perhaps the highest achievement of the dog is learning to discriminate between more or less flattened ellipses -- for unlike its sharp pitch-discrimination, based on native equipment, geometrical forms must represent the height of irrelevance in canine eyes. Yet once relevance has been revalued, and the perceptual analyser-devices have been established, the development of sharper sub-analysers or discriminatory filters must follow the stages outlined in Chapter X.*
In a seemingly casual aside, Hebb has remarked that 'the characteristic adult learning (outside of psychological laboratories) is learning that takes place in a few trials, or in one only'. [21] The implication is that the stamping-in of responses under artificial conditions in dogs, cats, or rats is quite uncharacteristic of the normal learning process. To try to base a human psychology on these procedures was a rather perverse approach.
Do Insects have Insight?
The work of Baerends, de Kruyt, Tinbergen, and Thorpe was rarely mentioned in the controversies to which I have referred, as if wasps, bees, fishes, and birds belonged to the fauna of another planet. As Thorpe remarked wistfully: 'Perhaps the arguments as to whether certain performances' of rats in mazes represent insight or trial-and-error learning would have been somewhat less prolonged if the abilities of some of the "lower animals", such as insects, had been known. . . . While it surprises no one that something like latent learning should be displayed by mammals and by birds with their proverbial powers of orientation, it may come as something of a shock to comparative psychologists who work primarily with mammals to find learning of this kind displayed at a high level among invertebrates. It is true that, with the confirmation of the work of von Frisch on the orientation of the hive-bee, we are now prepared to believe almost anything of bees, but there are certainly many insects other than bees, and many invertebrates other than insects, in which latent learning and similar performances can be discerned. The neglect of the study of invertebrate behaviour has given the impression that insight-learning is a characteristically human faculty hardly to be expected in a sub-primate mammal and, of course, out of the question in an arthropod. We now see what an astonishing misconception this is.' [22]