The Technique that we practise has been employed for over a hundred years. During that time much empirical knowledge had been accumulated about behaviour in general and the postural mechanisms in particular, but Alexander's original findings have never been contradicted.

However, as the whole subject has attracted world-wide interest, many teachers have felt a need to seek greater scientific knowledge in order to be able to assert 'scientific confirmation' of their methods.

Dr T. D. M. Roberts, formerly reader in Physiology at the University of Glasgow, is the leading authority in his field, having written the standard work on the Neurophysiology of Postural Mechanisms and also, a more popular work, Understanding Balance. In this latter work he asserts:

Everything to do with balance [comes] down to a problem of recognition. For example, if one is to make corrective movements to avoid overbalancing, one must be able, on some level, to recognise that the moment has arrived at which such corrective movements are called for.1

Dr Roberts goes on to argue that although animals and men evidently do regulate their posture, we must reject the idea, commonly entertained, that upright posture is maintained by some automatic regulating mechanism.

The practical teaching of all skills must be fundamentally empirical, and there needs to be recognised a clear distinction between practice and theory. But when it comes to putting forward explanations of theory, these must be correct and soundly based on the best scientific authority. If, as Alexander teachers, we appear to be scientifically illiterate, we cannot expect to engender much confidence in our pupils.

The actual words we use are important. Dr Roberts gave a lecture on this topic at the Constructive Teaching Centre in London in 1997. By publishing it in DIRECTION, it is hoped to bring it to the notice of a wider audience. His lecture should be studied with care, and although at first it may seem difficult for the non-scientist, an effort towards comprehension will prove rewarding.

- Walter Carrington

The words 'habit' and 'skill' are well established in common speech. They denote related, but different, classes of behaviour. In the same context, the word 'reflex' has crept into use in recent years but the important distinctions to be made between the three terms have tended to become clouded by inappropriate use. This article is intended to clarify the meaningful usages of these terms and of others that are also relevant.

It has been known since the time of Stephen Hales (1730) that certain reactions can be obtained even from a decapitated frog provided that its spinal cord is intact. For example, if a hind toe of a decapitated frog is pinched, the limb is promptly withdrawn, by flexion at hip, knee and ankle. In 1833, Marshall Hall characterised such reactions as 'reflex' on the grounds that some effect of the stimulus must be passed along sensory nerves to the central nervous system and that it must there be 'reflected back', so to speak, along motor nerves to produce the observed response.

By the turn of the century, Sir Charles Sherrington and his co-workers had uncovered a large variety of responses all sharing the same characteristics of being involuntary and of being dependent on the intactness of the connections to the nervous system. The responses may involve skeletal muscle (as in the withdrawal reflex and the knee-jerk), smooth muscle (as in the pupil constriction that follows a sudden increase in retinal illumination), or a gland (as in the salivation that follows the introduction of meat juice into a dog's mouth). In each case, a specific treatment applied by the experimenter is followed by a stereotyped pattern of response. The linkage between the sensory and motor neurons involved in each such reflex and the consequent stereotyped natures both of the effective stimulus and of the response, appears to be dependent on the anatomical relations established in the developing embryo as a result of the action of inherited factors, insofar as similar behaviour is found in all members of a particular species.

At the time when reflexes were first being described, little was known of the details of neuronal transmission. Stimuli and responses were accordingly described in terms of the experimenter's manipulation of the environment in presenting a 'stimulus' and of the observations of the experimenter as to the nature of the 'response'. This has to be borne in mind when interpreting such expressions as 'the same stimulus' or 'the same response'. It is highly unlikely that the 'sameness' implied in these expressions applies also to the detailed spatiotemporal patterns of the activations of the neurons involved.

A factor that is often overlooked is that both stimuli and responses can vary in intensity. This has to be taken into account in assessing 'sameness'. Thus, while a response may be characterised as 'the production of saliva', it may be relevant to consider also the actual amount of saliva produced in any specific instance.

A useful additional concept has been developed in relation to variations in the intensity of stimuli, namely that of the 'threshold'. This expression is used where a certain range of intensities fails to produce any response at all, and another (higher) range of intensities is reliably successful. The borderline between the two ranges is then referred to as the 'threshold' level of intensity for the successful eliciting of the response. While the concept is readily acceptable, the precise level of the threshold in any specific case is not available for measurement, since all we can say is that some intensities succeed while others fail, thus bracketing the threshold without ever observing it directly. Indeed, it often happens that the threshold appears to shift about somewhat.

In addition to considerations of intensity, care must be exercised also in setting out just what is to constitute the full set of conditions to be included in the specification of the stimulus. It may be necessary to mention other factors, such as the state of the animal, in addition to the details of the treatment to be applied by the experimenter. For instance, the knee-jerk reflex response is produced by the sudden increase in the tension in the patellar tendon when it is struck by the neurologist's hammer. However, the response does not occur if a similar sudden increase in tension results from the jerk arising on sudden loading of the limb during landing after a step. Furthermore, when a neurologist is testing for the integrity of the knee-jerk reflex, he arranges for the subject to be comfortably supported with the lower leg dangling free, because it has been found that these conditions are conducive to reliable assessment of the strength of the reflex response. The implication is that some process of recognition is carried out by the nervous system in discriminating those total situations in which a specific response is appropriate from other situations in which it is not.

Recognition processes of some kind occur even at the level of the individual neurons. The output activity of a neuron, in terms of the temporal pattern of impulses transmitted along its axonal branches, is dependent on the spatiotemporal pattern of impulses arriving at a very large number of synaptic sites distributed over the whole surface of its dendrites and soma, the activity at some sites being excitatory while at others it is inhibitory. Insofar as some patterns of input are successful in generating an output impulse while others are not, we may regard each neuron as performing a classificatory act, 'recognising' certain input patterns as appropriate to the generation of an output impulse and rejecting other patterns. There is a sense in which such recognition acts resemble the perceptual recognition of a 'gestalt', in that the response may be generated by any one of a class of patterns rather than only by presentation of a single complete set of specified constituent items.

These considerations apply to neurons of all types&emdash;sensory neurons, interneurons, and motoneurons. They also apply to the range of input patterns that are capable of generating a specific reflex response. Furthermore, a recognisably similar observable response may be produced by any one of a particular set of active motoneurons, since similar muscular tensions may be produced by various combinations of the motor units developing tension in a specific tendon.

While the motor patterns in reflex responses are more or less stereotyped, Pavlov's experiments on salivary reflexes in the dog indicate that the sensory pattern that triggers a specific response can be altered by the experimenter, using a special routine referred to as 'conditioning'. The resulting association between the new stimulus and the original response pattern is then referred to as a 'conditioned reflex'. For example, after establishing that the introduction of meat juice into a dog's mouth results in the reflex secretion of saliva, Pavlov, in a succession of trials, applied a previously indifferent stimulus, the sound of a bell, shortly before the presentation of the meat juice and found that, thereafter, the dog would salivate at the sound of the bell alone, without any meat juice being presented. In such conditioning experiments, the trigger gestalt is altered, while the response remains the same.

A chick pecking at cardboard demonstrates that begging behaviour
can elicited by test objects other than the head of a herring gull.

---

INNATE BEHAVIOUS PATTENS

The class of 'innate behaviour patterns', of which striking examples are found in insects and birds, shares some characteristics with the class of reflexes. In each case, there is: 1) a stereotyped trigger pattern, 2) a stereotyped pattern of behaviour, and 3) an inherited linkage between them.

However, the details of the output behaviour in the innate behavioural patterns often show considerable modification according to the nature of the environment in which the actions are performed. Thus, while the general architecture of a bird's nest is characteristic of the species, the details of the construction of each individual nest are dependent on the materials available to the actual nest builder. It seems therefore that, although the general form of the behaviour in these innate reactions is governed by inherited factors, there are considerable contributions to the detail of the execution that depend on voluntary control based on current sensory perceptions at the time of performance.

In certain cases, this detailed supervision of performance is absent, the resulting behaviour taking the form of a 'fixed action pattern'. One such example is the egg-rolling behaviour of the graylag goose. If an egg is removed from the nest of a sitting goose to a position a few inches outside the nest, the goose will reach down with its bill and roll the egg back into the nest. If the experimenter removes the egg after the rolling action has been started, instead of discontinuing the now irrelevant movement, the goose continues to move its head slowly backwards along the ground and into the nest as though rolling an invisible egg. Thus, once the action pattern has been initiated, it is persisted in right through to its conclusion in spite of the fact that the situation has changed and the movement is no longer appropriate.

A similar egg-rolling movement is initiated by other objects beside eggs, such as small cartons, including matchboxes or cigarette packets, provided that they roughly resemble eggs in a few particulars, such as approximate size. This brings out another feature characteristic of the innate behaviour patterns. In each case there is a specific sensory pattern that initiates the behaviour. Sometimes the constituents of the trigger pattern are surprisingly simple, in other cases they are exceedingly subtle and hard to specify.

An example of a simple trigger is seen in herring gull chicks. They beg for food by tapping their beaks against the beak of the parent bird as soon as it alights on the nest. It has been discovered that this characteristic form of begging behaviour can be elicited by test objects other than the head of a herring gull (see figure below), so long as they carry a red spot similar to that on the normal herring gull's beak. The test object can, in fact, be bizarrely different in shape from a gull's head, and gulls' heads that lack the red spot do not elicit the begging behaviour at all. The effective trigger stimulus is spoken of as an 'innate releasing mechanism', or 'releaser'.

More complex trigger patterns are involved in the various stages of courtship behaviour where each of the partners may perform a specific sequence of routines, each in turn triggered by the body language of the other partner. Innate behaviour patterns are involved in much of the courtship behaviour of humans as well as in that of other animals. In humans we can also see the sometimes disastrous consequences of confusion between complex releasers, as when an invitation to play is interpreted as an unwelcome courtship signal.

The effectiveness of certain releasers is often markedly different at different times. This leads to the concept of 'internal drives', such as hunger, thirst, or sexual appetite, which govern the times when specific releasers will be effective. In addition to the class of positive releasers, which initiate behaviour of particular patterns, there are also releasers which bring such behaviour to a termination. These could be labelled 'satiety releasers' that indicate that the current drive has been satisfied.

During a complex process such as nest-building, several different stages must be completed in the appropriate sequence: the search for a suitable site, the search for the right kind of building material, the choice, collection and transporting of the items of material back to the nest, and the intricate process of interweaving each item into the existing structure of the growing nest. Then there will be the changeover from collecting structural materials to collecting softer materials for the lining, and finally the eventual discontinuance of all collecting activity prior to the female's settling down to lay. Each stage is initiated and terminated by the appropriate releasers.

Reflexes and innate behaviour patterns are alike in that each depends on an inherited linkage between a specific trigger-recognition mechanism and the general form of the response pattern. However, the trigger for a reflex response, the 'adequate stimulus', is comparatively simply described and correspondingly is available for conditioning, while the releasers for the innate behaviour patterns can be very complex and do not lend themselves to conditioning. On the response side, reflex responses are highly stereotyped and clearly involuntary in nature, while the innate behaviour patterns may incorporate important components of voluntary action, superposed on a stereotyped general plan.

'Accidental' means that the precipitating cause does not correspond to one of the patterns in the definitions of the other classes. Sometimes it may be difficult to see that what happens 'accidentally' is part of a behaviour&emdash;as when a voluntary action 'accidentally' knocks something over. The overbalancing of the bdy that occurs in consequence of the short time-course of the forces generated by individual motor units may be classed as accidental, even though the consequent changes in stress patterns may initiate responses classifiable under the other headings.

The stereotyped nature of the responses in reflex and other innate behaviour patterns implies that, at some point in the evolutionary history of the species, some survival advantage has been conferred by specific patterns of motor performance in contexts where the corresponding adequate stimuli and releasers have been recognisable. The necessary recognition processes can be accounted for in terms of the adaptive behaviour of large assemblages of neurons, each having multiple inputs and multiple output pathways, since comparable recognition acts can be seen to be performed also by man-made large assemblages of neuronal analogs consisting either of pieces of hardware or of equivalent computer simulations with correspondingly pseudo-random interconnections. A requirement of such computers is the provision of some sort of 'reward function' that will guide the adjustment of the weighting, or effectiveness, of the transmission across each of the various nodes in the network of interconnections. In animals the necessary reward function is available in the form of the satiety releasers related to each of the internal drives.

A consequence of the highly stereotyped responses available in specific reflexes is that the testing for the presence or absence of certain reflexes is of use in practical neurological diagnosis, because the absence of a particular reflex in a patient implies that certain specific nervous pathways in that patient must have been compromised in some way. For this reason there is considerable utility in restricting the use of the term to genuine examples of reflex activity and in avoiding the nowadays all too common loose usage of the term to cover any reasonably rapid action, such as many of the actions of the trained athlete.

© Francois Le Diascorn 1983

Each animal has a propensity to modify its behaviour
according to the rewards it receives. An experimenter can make use
of this propensity to train animals to perform specific tasks.

---

LEARNING

The ability of the trained athlete to react quickly to certain situations is a result of learning rather than of reflexes, where the expression:

learning&emdash;implies the modification of behaviour in the light of experience; and

experience&emdash;is the totality of the sensory input to the central nervous system of the individual.

Behaviour first manifests itself in the activation of the motoneurons controlling the muscles that produce the movements visible to the onlooker. The movements affect various sense organs which send their sensory messages to the nervous system. The messages activating the motoneurons are also monitored, so that the nervous system has access to information both about the pattern of motoneuron activation and about its consequences. The apparently random waving movements of the limbs of a new-born infant generate, not only information about the movements themselves, but also about encounters between the limbs and other objects in the environment. Some of these encounters will lead to the activation of reward processes associated with exploration, and in this way the infant learns about the nature of its environment and the way the environment interacts with its own behaviour.

It is a common property of animals that, if they associate some improvement in their situation with their performance of a specific behavioural act, they thereafter tend to perform that action more frequently. This propensity is at the heart of the learning process, and forms the basis for various training routines.

An experimenter can train an animal to perform certain specific actions by using the propensity which each animal possesses to modify its own behaviour according to the rewards received. The first step is that the experimenter must select some procedure by which he can affect the animal's environment in a way that is advantageous to the pupil. An obvious and much used such procedure in animal work is the presentation of a morsel of food. The experimenter then studies the pupil's apparently spontaneous behaviour to identify some particular action that can be used as a basis for a training routine. The next stage is to present a food morsel promptly every time the pupil performs the selected action.

This procedure is known as 'positive reinforcement', and the selected behaviour comes to be performed frequently, so long as the reward continues to retain its effectiveness. The training is then extended by requiring that additional, previously spontaneous, actions are performed in conjunction with the act originally selected, thus 'shaping' the pupil's behavioural repertoire. By proceeding stepwise in this fashion it is possible to produce quite marked changes in the pupil's behaviour.

The training routine can be automated using a device that presents the reward mechanically when the pupil makes some move that triggers a suitable detector. This leads to a description of the training process as 'operant conditioning', where the 'operant' is the animal that performs the action to be rewarded (operates a lever or whatever), and whose behaviour is changed ('conditioned') by the reinforcement provided by consistent reward.

The use of the word 'conditioning' can lead to confusion. This procedure for operant conditioning is not the same as that used to set up 'conditioned reflex responses'. In 'classical conditioning', the experimenter's intervention (with the sound of a bell) has to precede the presentation of the adequate stimulus (meat juice in the dog's mouth) so that, after many repetitions, a response of the original form (salivation) is produced in response to the signal (bell) without the unconditioned stimulus (meat juice) being presented to the animal.

The possibility of confusion with routines for establishing conditioned reflexes becomes particularly insidious in relation to the use of positive reinforcement to train an animal to perform a specific action on command. In training to a cue, positive reinforcement by reward is first established, as described in the previous section. The presentation of the rewards is now associated with some advance signal to which the animal would otherwise be indifferent. After a number of repetitions of the combination of signal and reward, the consistency of rewarding is altered. Instead of rewarding every instance of the animal's performance of the selected action, the reward is presented only on certain occasions, and then only in conjunction with the signal cue. After a time, the animal comes to associate the signal with the expectation of receiving a reward. The frequency with which it performs the selected act in the absence of the signal now declines, as such instances are no longer rewarded, and the animal comes to perform the action promptly on cue and not at other times.

The cue to produce a response in a conditioned reflex has to precede the unconditioned stimulus. In operant conditioning on the other hand, the experimenter's intervention (reward) has to follow the action to be reinforced. The response in the conditioned reflex is the same as the response in the unconditioned reflex on which it is based, whereas the response in operant conditioning can be of any form and it does not necessarily conform to the pattern of response that can be elicited in any reflex. In view of these differences, it is clearly incorrect to assert that learning processes in general are based on conditioned reflexes.

The promptness of trained responses can sometimes be encouraged by 'negative reinforcement'. This term is used for routines in which, if the animal fails to perform the desired action promptly in response to the cue, some additional treatment is applied that the animal is to understand as a rebuke, or at any rate that puts it into a situation it would prefer to avoid.

Negative reinforcement routines, in the form of punitive sanctions of one kind or another, are very widely advocated as the preferred option for modifying undesirable human behaviour. The outcome, however, is all too often very far from what was intended. The reason for this is that the person whose behaviour we wish to change may discover alternative ways of escaping from the punishment situation. The resulting new behaviour may be just as undesirable, from our point of view, as the pattern we originally set out to alter, but this new pattern itself becomes reinforced by the rewarding effect of the escape from punishment. Further punishment just makes matters worse. Because of this inherent disadvantage of the negative reinforcement strategy, it is important not to neglect the proven efficacy of positive reward and to recall that in many cases a mild rebuke can be more effective than downright punishment.

The modification of behaviour in the light of experience goes on throughout the life of the individual. Much of an animal's learned behaviour serves to make changes in its environment. Some of these changes are particularly rewarding, and the animal learns to adjust its behaviour at repeated trials to become more and more expert in producing those changes in its environment that are especially desirable. The expert performance of actions that are highly appropriate to particular situations is what we refer to as a 'skill'. Continuous close scrutiny is required at each successive stage in a skilled action, together with monitoring of the effects produced. A repertoire of learned component movements is developed, each appropriate to a particular stage reached during the performance of the overall task and from which the currently appropriate component is selected in turn as occasion demands.

Some situations occur frequently in the life of an animal and, although skilled responses are called for, the proficiency acquired from repeated rehearsals allows the intensity of continuous close scrutiny to be relaxed. Eventually such actions come to be performed, when appropriate, without conscious supervision, and even the fact that the behaviour has been invoked may fail to be registered in consciousness. An action that has reached this level of development is referred to as a 'habit'. Many of the components of balancing behaviour and of locomotion, as actually performed in practice, turn out to fall into this class, although reflexes producing related movements have also been identified.

One may often read of an animal, or a person, being 'conditioned to behave in a particular way' where all that is intended is that the subject has 'learned to do it'. The usage ignores the fact that conditioning is done by someone other than the subject, while learning is something that the subject does for himself. Furthermore, a good deal of learning takes place without the intervention of any outside party.

There is a very important class of learned actions that closely resemble the responses in certain ordinary reflexes. These are the 'anticipatory pre-emptive actions', which are performed in situations where the ordinary reflexes are likely to be evoked but where the pre-emptive actions are set off, before the thresholds are reached for any reflex response, by expectations rather than by direct stimuli. The effect is that corrective actions, for example in the avoidance of overbalancing, can be initiated early, before the situation has become so extreme as to trigger off the ordinary reflexes of balance. The advantage of these 'anticipatory pre-emptive actions' is that they avoid the delays that are inevitably incurred in the conduction of nervous messages in the ordinary reflexes. This improved promptness in responding can be particularly important for the maintenance of balance, since it leads to a much smoother and more secure performance. The individual limb movements in locomotion are all pre-emptive, rather than reflex. This allows scope for changes in speed, in gait, or in direction, and even for the aiming of individual footfalls in the avoidance of obstacles.

The anticipatory pre-emptive actions, once developed, rapidly become habitual; that is to say, the reaction comes to be produced almost automatically whenever the 'trigger gestalt' is detected. In the case of the reactions to the imminence of overbalancing, the trigger gestalt for the initiation of pre-emptive correcting actions presumably includes the detection of a developing trend in the relation between the line of action of the resultant supporting thrust and the perimeter of the available area of support.

Habits that are performed without the subject being aware of the precipitating events are particularly difficult to alter. The recognition of the triggering gestalt often occurs before one has become aware of the changing situation. That is why we are so seldom able to say what it was that triggered off a particular habitual movement. We just find ourselves performing the habitual action 'without thinking'. It is this feature of habits that makes it so hard to change them. It also highlights the importance of avoiding the development of bad habits.

In the attempt to change a person's habit, two things appear to be essential. The first is that the pupil has to have developed an adequate desire to change and, to achieve this, the instructor needs to use his ingenuity to build up a suitable degree of motivation in his pupil. The second requirement, and one which is much more difficult to meet, is that the person with the habit has to be brought to a condition in which he is aware of the 'feel' of that changing situation which will eventually produce the habitual action. It is only when he has learned to recognise this crucial moment that he can take effective action to produce an alternative pattern of behaviour.

In the training of an animal, one cannot suggest that the animal should watch out for a particular set of conditions. It falls to the trainer to spot the crucial moment at which it might be appropriate to intervene. He needs a finely-tuned awareness of what the animal is doing, based on a close attention to the most minute movements that he makes, so as to be able to anticipate that he is about to manifest the habitual behaviour that is to be changed. It is only when the trainer gets the timing right, and intervenes with the appropriate action, that he has any chance of successfully modifying an established habit. However, if one can avoid the situations in which a bad habit is likely to be performed, the force of that habit will gradually dwindle through lack of the reinforcement of repetition.

"The correction of some postural habits
can significantly improve a person's well-being."
(FM Alexander teaching)

---

IMPLICATIONS FOR ALEXANDER TEACHERS

The correction of some postural habits can significantly improve a person's well-being. For example, many people habitually exert additional effort to supplement the muscular actions called for in maintaining uprightness and in supporting the limbs. As this extra effort frequently involves simultaneous activity in opposing muscle groups, the effect is to stiffen the joints and to cause undue fatigue and even pain. To correct such habitual inappropriate muscular activity, the therapist must somehow induce the patient to become aware of the proprioceptive signals that initiate the unwanted activity. If this awareness can be achieved, the patient is provided with an opportunity of voluntarily refraining from initiating the stiffening actions. His posture thereafter becomes more supple and, consequently, more comfortable to maintain.

- Dr. Tristian D. Roberts

ENDNOTES

1. Roberts, T. D. M., Understanding Balance: the Mechanics of Posture & Locomotion; Chapman & Hall: London (1995) p.ix (Preface)

ABOUT THE WRITER

Dr. Tris Roberts is a scientist specialising in balance, posture and locomotion. He is a Senior Research Fellow, formerly Reader in Physiology, at the University of Glasgow, Scotland. He is the author of Neurophysiology of Postural Mechanisms (2nd Edn. London: Butterworth, 1978), Understanding Balance, (London: Chapman & Hall, 1995), and Equestrian Technique, (London: J.A.Allen, 1992). In 1984 he gave a weekend seminar in Glasgow on "Basic Mechanisms of Posture" to Teachers of the Alexander Technique from all over Britain.

11 Menzies Drive
Fintry, Glasgow G63 0YG
Scotland.

Email - gpaa30@udcf.gla.ac.uk