Summary: Rigidity and Freedom
Let me recapitulate some points which emerge when the observations in the present chapter are taken in conjunction with the broader issues discussed in Book One.
On the elementary levels of learning a skill a varying amount of stamping-in is required, depending on she organism's 'ripeness' for the task; or, to put it the other way round, depending on the 'naturalness' of the task relative to the organism's existing skills. Learning to type requires more stamping-in than learning to ride a bicycle; the former is comparable to the blindfold memorizing of a maze, the latter to the gradual adjustment of various interlocking servo-mechanisms. In both cases the learning process consists in the integration of elementary skills -- the members of the nascent matrix -- into a single pattern which can be activated as a unit. But even in acquiring a mechanical skill like typing, bit-by-bit learning plays in fact a lesser part than seems to be the case. The typist's mental map of the keyboard is' not simply a rote-learned aggregation of twenty-six letters (plus numbers and signs) distributed at random; it is a 'coded' map, structured by a system of co-ordinates -- the resting position of the fingers -- and by the frequency-rating of letters, syllables, etc. These patterns, superimposed on the keyboard map, could be compared to the mnemonic aids used in the learning of nonsense syllables. Whole-learning invades bit-learning at every opportunity; if the meaningless is to be retained, the mind must smuggle meaning into it.
Once a skill has been mastered so that it can be activated as a trait it functions more or less autonomously and automatically. This applies to both perceptual and motor skills, from the perceptual constancies and motor reflexes upwards. Learning to find the right key on the keyboard requires concentration, focal awareness; but when the letter-habit has been acquired it becomes 'instinctive', unconscious; attention is freed to concentrate on meaning, and can 'let the fingers take care of themselves'; their control is relegated to lower levels of awareness and, in all likelihood, to lower levels in the nervous system. Thus the work of Gastaut and Beck clearly suggests that 'once we have learned something we no longer rely so much on our cortex and reticular formation. Those things we do without thinking . . . may depend more on the older primitive parts of the nervous system such as the limbic structures, thus releasing higher centres such as the cortex for other tasks. . . . Common sense indicates such a possibility.; electro-physiology suggests it. . . .' [5]*
The same skilled action -- driving a motor-car or playing a nocturne by heart -- can be performed automatically, or in semi-conscious absent-mindedness, or with full concentration. But the motorist who concentrates on driving fast along a crowded road has his attention focussed on matters of general strategy -- e.g. whether it is safe to overtake or not, whereas the actual manipulation of the wheel and pedals are still carried out automatically; and the pianist trying to give his best, still finds the keys automatically. We again find confirmed that the code which controls skilled behaviour always operates through sub-codes which function on lower levels of awareness. Shifting the focus of attention to these sub-codes produces the familiar 'paradox of the centipede'. Its equivalent in perception is the loss of meaning which results when a word is repeated monotonously and attention becomes focussed on the Klangbild (cf. 'ce-du, ce-du, ce-du', Book One, p. 75 f.; even more painful is the semantic paralysis which sometimes befalls a writer while correcting the proofs of a forthcoming book.
The lower we descend in the hierarchy the more stereotyped, reflex-like activities we find; and vice versa, flexibility increases with each step upward. The more complex the skill, the more alternative variations it offers for adaptable strategies: a matrix on the n+1 level has more degrees of freedom than a matrix on the n level. But whether they will be utilized and produce varied performance, depends on the environment. Monotonous environments induce repetitive, stereotyped habits; the degrees of freedom in the matrix freeze up. 'Overlearning' is the fixation, through repetition in unvarying conditions, of one among many possible variations in the exercise of a skill at the expense of all others. Thus habits become automatized (a) because they operate on the lower strata of the hierarchy with few degrees of freedom, like hitting a typewriter key or depressing the accelerator pedal; (b) when a complex skill is reduced through environmental monotony to a single-track habit. 'Monotony' is of course a subjective term referring to lack of change in those features of the environment which are relevant to the subject's interests. For all we know the streets of Koenigsberg through which Emmanuel Kant took his fixed walk at a fixed hour for forty years might have been wildly exciting to another person.
The integration of motor-patterns into larger and more complex skills in the process of learning is paralleled by a similar progression on the perceptual side. The telegraphist who has advanced from 'letter-habits' through 'word-habits' to 'phrase-habits' in his sending technique, has at the same time learned to take in several words and even phrases 'at a mouthful'. The pianist takes in a whole musical phrase from the score at a glance; both input and output are no longer measured in bits but in chunks.* The more complex the skill, the bigger the chunks in space or time which must be taken into account. The skilled soccer player keeps his eye on the ball, but is at the same time aware of the positions and peculiarities of the other players on the field. The motorist, driving to his office, chooses the least congested road among several alternatives by consulting the mental map in his head. The typist, who deliberately lags a phrase or two behind dictation, expands the duration of the psychological present to take in a bigger chunk of meaning. While listening to speech or music we do the same; while talking we trigger off long sequences of muscular patterns as a whole. As we become more proficient in any skilled activity, we learn 'to put feedback loops around larger and larger segments of our behaviour'. [6]
Though motor learning proceeds, generally speaking, from lower to higher levels, and performance in the reverse direction, this does not mean that in performing we run through the whole gamut of the learning process in reverse gear. As one learns to play a sonata by heart, one needs less and less often to consult the score, and in the end the visual feedback which was indispensable during learning can be dispensed with entirely; the habit now functions autonomously. The skilled pianist can play blindfold, a man can knot his tie without looking into the mirror, the physician can tell the patient's pulse without looking at his watch, the adult reads without spelling out the letters. When the skill has been mastered, the props which served the learning process are kicked away -- as Maxwell kicked away the scaffolding of his mechanical model when he arrived at his equations (see Appendix I). In this respect, too, the learning process is irreversible.
The autonomy of the codes which pattern behaviour is a phenomenon which we have met on all levels -- from the self-regulatory activities of the morphogenetic field, through the fixed action-patterns of instinct behaviour, to the perceptual frames responsible for constancies, illusions, and our ways of seeing the world through coloured filters, as it were. But on the level of complex skills, the 'self-assertive' tendencies of acquired motor-patterns are particularly striking. To repeat an obvious example, one cannot disguise one's handwriting sufficiently to fool the expert; even the skilled burglar has his individual style in safe-breaking which gives him away. Autonomy and self-government are basic principles in the hierarchy of skills. Thus 'the performance 'of very quick movements ', Lashley observed, 'indicates their independence of current control. "Whipsnapping" movements of the hand can be regulated in extent, yet the entire movement, from initiation to completion requires less than the reaction time for a tactile or kinaesthetic stimulation of the arm, which is about one-eighth of a second, even when no discrimination is involved. . . . The finger-strokes of a musician may reach sixteen per second in passages which call for a definite and changing order of successive finger-movements. The succession of movements is too quick even for visual reaction time. . . . Sensory control of movement seems to be ruled out in such acts.' [7]
Similar conclusions were reached, as already mentioned, by Ruch, concerning voluntary movement in general. In view of the rapidity of skilled movements which are too fast to leave room for visual or proprioceptive feedback control, Ruch, like Lashley, assumed the operation of pre-set time-tension patterns of muscle contraction in the nervous system: 'The cerebral-cerebellar circuit may represent not so much an error-correcting device as a part of a mechanism by which an instantaneous order can be extended in time . . . and thus reduce the troublesome transients involved in the correction of movement by output-informed feedbacks.' [8]
The tendency to reduce those 'troublesome' feedbacks to a minimum is the essence of habit-formation and automatization. It follows the principle of parsimony; if we had to concentrate on each movement we made, there would be no room for thought. On the other hand, this inherent tendency to form neural organizations which, one might say, jealously defend their autonomy against interference from a changing outside world, makes us all, in varying degrees, the slaves of habit. We may reduce the degree of enslavement, but the basic predicament is inherent in the hierarchic structure or nervous organization, where 'the structure of the input does not produce the structure of the output, but merely modifies intrinsic nervous activities that have a structural organization of their own'. The quotation (repeated from p. 434) referred to instinct behaviour and the lower motor functions, but it is equally applicable, as we have seen, to complex, acquired skills. These may have a high degree of flexibility, but they nevertheless operate through automatized sub-skills on the lower ranges of the hierarchy, which manifest themselves in the individual 'touch' of the pianist, the 'style' of the tennis-player, the fixed mannerisms, quirks, idiosyncrasies, and unconscious rituals which are our personal hallmarks.
How much of his potential freedom a person puts to active use depends partly on environmental factors -- the novelty, intensity, vexatiousness, etc., of the stimuli to which he is exposed. But the nature and amount of stimutation derived from a given input depends, of course, on personality stricture. One type of individual will respond to monotonous situations with stereotyped reactions; another type will find monotony vexatious, that is to say, stimulating. 'Active boredom', as this kind of reaction may be called, can provide alternatives to habituation; the subject may experience the very absence of change as a novelty -- as prison is a novelty to the first offender; and since the environment refuses to offer variety, he will vary his own performance to provide it. Hence the apparently spontaneous changes in fashions and crazes, not only in human society but also in colonies of captive chimpanzees.
If, on the other hand, the challenge from the enviromnent exceeds a critical limit, behaviour will either become distintegrated, or the challenge will be met by an original, 'super-flexible' response -- a restructuring of the pattern of the skill. We have met examples of this on all levels, from the 'prenatal skills' of morphogenesis, through Bethe's mutilated insects to Lashley's rats and Köhler's chimpanzees. The complex, acquired motor-skills, which we discussed in this chapter, are capable of equally impressive emergency-reorganizations. The first aircraftsman who, when his brakes refused to function on landing, saved his plane by opening his parachute through the rear-window, achieved a true bisociation of two unconnected skills. The violinist who finishes his piece in spite of a broken E string; the typist managing on a half-broken machine; the secret tunnel-builders in prisoner-of-war camps; the legless war pilot winning a Victoria Cross; Renoir, crippled with arthritis, continuing to paint with a brush fixed to his forearm -- they all gave proof of an unexpected, creative surplus-potential in the nervous system. Such accomplishments are more impressive than the quasi-miraculous feats performed in panic or rage -- the latter are of a quantitative order and do not involve the reorganization of pattern.
The homologous nature of the basic principles which operate on different levels of the hierarchy becomes evident when we remember the conclusions which emerged from the discussion of instinct behaviour: 'At one end of the scale we find rituals, fixed action-patterns, vacuum and displacement activities -- rigid, automatized, and compulsive, petrified habits. At the other extreme we find . . . codes which govern behaviour of remarkable flexibility, and original adaptations which lie outside the animal's normal skills and habit repertory. [10]
NOTES
To p. 549. See, for instance, the rather desperate footnote on p. 197 of Miller et al (1960): 'One reason for much of the trouble on reaching an agreement about the way the brain works was that two of the authors stubbornly persisted in trying to talk about it in terms appropriate to the dry hardware of modern digital computers, whereas the third was equally persistent in using language appropriate to the wet software that lives inside the skull. After a decade of cybernetics you might think the translation from one of these languages into the other would be fairly simple, but that was not the case. The relation between computers and brains was a battle the authors fought with one another until the exasperation became unbearable.'
To p. 550. Experiments by S. M. Evans (New Scientist, 2 May, 1963) have shown that the supraoesophageal ganglia of the ring-worm (its 'brain') are essential for learning but not for memory storage since learned habits will be retained after removal of the 'brain'. Once the habit is acquired, it is apparently transferred 'to a storage centre which is presumably somewhere else in the nervous system'.
To. p. 551. Cf. also: 'in rapid sight reading it is impossible to read the individual notes of an arpeggio. The notes must be seen in groups, and it is actually easier to read chords seen simultaneously and to translate them into temporal sequence than to read successive notes in the arpeggio as usually written' (Lashley in the Hixon Symposium, p. 123).
XII
THE PITFALLS OF LEARNING THEORY
A Glance in Retrospect
In the course of the past fifty years, learning theory has been one of the central battlefields of psychology. 'One may say broadly', Bertrand Russell wrote in 1927, 'that all the animals that have been carefully observed have behaved so as to confirm the philosophy in which the observer believed before his observations began. Nay, more, they have all displayed the national characteristics of the observer. Animals studied by Americans rush about frantically, with an incredible display of hustle and pep, and at last achieve the desired result by chance. Animals observed by Germans sit still and think, and at last evolve the situation out of their inner consciousness. To the plain man, such as the present writer, this situation is discouraging. I observe, however, that the type of problem which a man naturally sets to an animal depends upon his own philosophy, and that this probably accounts for the differences in the results.' [1]
Russell's remarks remain true, even though some eminent psychologists deny that they have a philosophy at all and hold all theory-making to be 'wasteful and misleading'. [2] Not only the choice of problem, but also the choice of animals is characteristic of the experimenter's bias. Köhler, desirous to prove insight and intelligence, concentrated mainly on chimpanzees. Skinner's best-known books are called The Behaviour of Organisms (1938) and Science and Human Behaviour (1953); but -- as Hilgard said -- 'neither title betrays that the precise data were derived largely from experiments on rats and pigeons. It is somewhat anomalous that a systematist who refuses to predict what a rat or pigeon will do -- because such prediction does not belong in a scientific study of behaviour -- is willing to make confident assertions about the most complex forms of human behaviour, economic, political, religious'. [3]
Hull's attempt to create an all-embracing theory of behaviour was almost entirely based on the bar-pressing activities of rats. This was considered by Hull and his school as a sufficiently solid basis to derive from them 'the basic laws of behaviour . . . including the social behaviour of man'. [4]
Lastly, the German school of ethologists -- Lorenz, Tinbergen, etc. -- concentrated mainly on highly ritualized forms of animal life in birds and insects. Thus each school developed its special universe of discourse, moving in a clo
sed system, concentrating on their favourite animals in their favourite experimental situations -- dogs dripping saliva througli fistulae, cats raging in puzzle-boxes, rats running through mazes, geese being 'imprinted' by Dr. Lorenz ambling on all-fours.
But the data from these highly specialized, experimental trends did not add up to a coherent picture, and each school had a tendency to ignore what the others were doing. Thus, for instance, in Skinner's "Science and Human Behaviour", which was intended as a textbook, the index contains neither the word 'insight', nor the names of Köhler, Koffka, Lewin, Tolman, Hull, Lashley, or Lorenz; only Thorndike and Pavlov are mentioned by name as theorists of learning with some merit. And vice versa, I have searched in vain for the name of Pavlov in the indices of Köhler's and Koffka's books. [5] Thus much of the controversies in learning theory resembled less a battle than a game of blind man's buff.
How, just at a time when the mechanistic conceptions of the nineteenth century had been abandoned in all branches of science, from physics to embryology; how just at that time, in the 1920s, the concept of man as a rigid mechanism of chained reflexes could become fashionable in cultures as different as the United States and the Soviet Union is a fascinating problem for the historian of science. The Pavlovian school in Russia, and the Watsonian brand of Behaviourism in America, were the twentieth-century postscript to the nineteenth century's mechanistic materialism, its belated and most consistent attempt to describe living organisms in terms of machine theory.
The Act of Creation Page 66