PERFORMANCE INFORMATION FEEDBACK - A NECESSARY INGREDIENT FOR SKILL LEARNING
[Extracted from Rushall, B. S., & Siedentop, D. (1972). The development and control of behavior in sport and physical education. Philadelphia, PA: Lea & Febiger. (pp. 55-63)]
A consequence that is of great importance in the acquisition and maintenance of motor behaviors is information feedback (IF). IF has been defined as information resulting from a response (Fitts and Posner, 1967). The theoretical function of IF in skilled performance is still a matter of dispute. Skinner (1969) has recently suggested that IF should not be used as a synonym for operant reinforcement. Learning theorists, on the other hand, have begun to pay particular attention to the motivational properties of IF (Fitts and Posner, 1967; Anokhin, 1969; E. A. Bilodeau, 1969). A discussion of the feedback/reinforcement properties of performance information is presented in Chapter 10. The reader is encouraged to take this discussion into account when examining the more practical features of IF presented in this section. Coaches and teachers have the power to control at least some features of IF. These features act as secondary reinforcers.
Theorists are in complete agreement that IF is essential for skilled performance. After reviewing a number of studies that manipulated various sets of feedback factors, Bilodeau concluded the following about the frequency of information feedback.
"This appears very clearly in the need for IF in improving and sustaining performance, or in three repeatedly demonstrated empirical effects: performance fails to improve unless IF is introduced; performance improves with IF; and performance either deteriorates if IF is withdrawn, or shows no further improvement." (I. McD. Bilodeau, 1969, p. 260.)
Such a conclusion is, in itself, of importance to physical educators as it suggests that learning environments in sport and physical education can be improved by focusing on the frequency and clarity of IF.
A variety of descriptions exist about the forms of IF derived from performance of a skilled act. For purposes of this text, IF will be categorized using the framework suggested by Holding (1965).
There are two main types of IF: intrinsic and artificial. Intrinsic IF refers to information that is inherent in a task. In tennis, for example, the lines on the court provide information as to whether a shot was good. The net and markings of the service area provide information as to the efficacy of a serve. The primary role of intrinsic feedback is that it allows the performer to evaluate a response. It provides a frame of reference so that errors in response can be detected and attempts made to correct them. It is this discrepancy (error) between the actual response and the desired response that acts as input for the next response, and it is in this sense that theorists talk about feedback loops. It is obvious that some sports activities provide a great deal of intrinsic evaluative feedback. This is true of tennis, bowling, badminton, and golf for example. Others such as track running events, swimming, and dance provide very little.
Intrinsic IF refers also to the normal organic sensations which occur during and following a physical performance. These sensations arise through kinesthetic stimulation and provide cues about the rate and location of movements. They are important for attempts at replicating successful responses. Typically, when a response achieves a desired end the performer attempts to "do the same again." Kinesthetic IF provides cues which indicate to the performer the amount and type of movement that has been performed.
Although kinesthetic IF is an important variable in most motor learning theories it is difficult for the teacher or coach to use. It is potentially available through manual manipulation of the performer by the teacher. When the limbs and bodily positions of a performer can be manipulated by the teacher, some form of kinesthetic IF is produced. The idea, quite simply, is to have the learner experience the "feel" of proper form. Beginning swimmers can be handled in the initial stages by having the instructor rotate their arms, turn their heads for breathing, hold them in a horizontal floating position, and move their legs in a flutter kick. Young children who lack the facility for verbal comprehension are also aided through manual manipulation. Lawther (1968) suggested that this is also a useful technique when instructing aged or handicapped performers.
Certain kinds of training devices also create situations in which a learner may experience the kinesthetic feedback associated with an acceptable form of movement. The safety belt in gymnastics allows the learner to experience the kinesthetic IF from rotary movement without the fear of disastrous consequences. Basketball teachers have for years used "blinders" as an aid to teaching higher levels of skill in dribbling. The blinders prevent the learner from looking down, thus minimizing the visual IF of seeing the hand control the ball. With the visual IF minimized, the learner is forced to attend to the kinesthetic IF, thus developing the ability to dribble by "touch."
Manual manipulation is obviously of limited usefulness simply because only one performer can be manipulated at a time, thus making the technique inefficient in terms of time-cost. It seems to be most useful with normal learners when other methods have been attempted and have not produced the desired results. Training devices which require the performer to approximate a desired topography (form) offer a much more fruitful approach to the problem of kinesthetic IF. It should be hoped that physical educators will experiment with various kinds of training devices especially in those activities where the amount of intrinsic evaluative IF is minimal.
Practices already exist in coaching and teaching skills which can provide the performer with a "feel" for a particular action. This "feel" refers to the kind of kinesthetic IF which is associated with the final desired level of skill. The method for providing this IF has several characteristics:
These features will be discussed more fully in Chapter 7. [See sections on "shaping" in this issue of Coaching Science Abstracts.]
Artificial IF refers to information that is not usually available in the performance of a task. This is extra information that is added by the teacher for training purposes. It is also often referred to as augmented feedback. The most common form of artificial IF is evaluative comments by an instructor. Whenever a teacher or coach suggests a change in form or comments on a particular response, the information provided comes under the classification of artificial IF. While being of importance in many teaching situations, the limitations of verbal IF are obvious. Feedback can be supplied to only one learner at a time and the teacher's attention must be focused on the one learner for a minimal period of time in order to evaluate the performance.
The potential importance of artificial IF for teaching in physical education and sports is in finding ways of providing extra performance information to many learners during or immediately after their performances. As such, the use of pacing machines, ergometers, metering devices, and other evaluative tools merits a great deal of investigation.
Results of research dealing with the effectiveness of artificial IF have not been too encouraging. Many studies find that performance is increased while the artificial IF is present but deteriorates once it is removed (E. A. Bilodeau, 1969). This kind of result has been particularly prominent in motor skills research. Research in other fields has yielded more optimistic results for the potential of artificial IF (Collier, 1971). Holding (1965) has spoken quite succinctly to the problem of the permanency of IF.
The difficulty about putting in artificial knowledge of results is that its effects may not last after its removal. Eventually the learner must come to rely upon the intrinsic cues. There is no point in learning to rely upon information which will not be there when training is finished. The success of techniques of augmenting feedback will depend upon whether they call attention to the intrinsic cues or make possible control of the relevant responses in a way which can later be taken over by the intrinsic cues.
It appears that the task of the teacher and coach is to find forms of artificial IF that will increase performance during training and allow for transfer to intrinsic IF sources which will in turn maintain performance at high levels after training. The operant strategies that follow in later chapters address this problem directly.
A second level at which IF types may be categorized is concurrent or terminal. Concurrent IF refers to information that is available during the performance of a task, that is, while a response is being made. Intrinsic kinesthetic IF is of the concurrent variety as is the information provided by the speedometer on a car or the self-monitoring of the pace of a runner by a clock that is always within the visual field. Terminal IF refers to information that arises as a result of a completed response. It is often referred to as knowledge of results (KR). Arrows in a target and the success or failure of a jump shot are examples of terminal IF. It would appear that terminal IF is of great importance for the learning of individual skills such as the volleyball serve, tennis forehand, or football punt. Research has been quite consistent in revealing that terminal IF is not only important to the learning process, but, indeed, is a necessary condition for any learning to occur.
It would be beneficial if the higher performances developed with concurrent, artificial IF could be maintained after the training period. It would appear that the best method by which to accomplish this would be to gradually transfer control of the behavior from the artificial source to the intrinsic IF source in a manner similar to the procedure known as stimulus fading. Siedentop (in press) attempted to do this when having subjects learn to walk at a particular pace around an oval course. The "fade" group started out with full concurrent artificial IF, that is, the elapsed time from the start of each trial was communicated verbally to them. The frequency of the artificial verbal IF was gradually reduced until during the final training sessions the subjects received only terminal IF. Throughout the training period the faded IF was sufficient to maintain a very high level of performance, but during the transfer sessions when no IF was available this group did not perform as well as another group that had been trained under terminal IF conditions. This study demonstrated the difficulty of transferring artificial KR control to intrinsic IF cues. Methods need to be found for doing this. It seems to be possible in some situations (Rushall, 1970) but difficult in others.
A third level of categorization of IF should be mentioned briefly. This is immediate or delayed IF. With animal subjects IF must be reasonably immediate for learning to occur, but with human subjects terminal IF can be delayed somewhat without hindering the acquisition of skill (Annett, 1969). The same is not true, however, for concurrent feedback. Any recognizable delay in concurrent feedback virtually destroys the subject's ability to perform. This phenomenon is most dramatically displayed in experiments dealing with delay of concurrent feedback in speech. In such an experiment the performer most often stutters, slurs words, and acts generally as if inebriated (Lee, 1950).
Although there has been much research done in the area of feedback during the past decade many questions remain unanswered and a great deal of creative work must be done in applying the results of feedback research to the teaching of sports skills. Most importantly, what is known is that improvements in performance are directly related to the availability of IF, and when IF is removed performance most often deteriorates. It is also obvious that artificial IF will be important to the degree that it calls attention to intrinsic IF that will maintain the performance level after the artificial IF has been removed. There is also still much to be learned about the effects of different amounts and specificities of IF.
It is the task of the teacher and coach to apply the results of research intelligently and creatively. In many present situations in physical education and sports the level of IF is notoriously low, especially when one considers its basic importance in the learning process. Some activities do provide instant results. A jump shot is either successful or unsuccessful. The shooter is even able to see if his attempts are more closely approximating success. A golfer receives rather clear IF as does the archer, place-kicker, and bowler. Activities which are tests of accuracy generally provide enough IF for an individual to improve considerably over time without the aid of a coach. The reinforcement (IF and self-evaluated improvement) gained from performance is sufficient to maintain the practice behavior of an individual for long periods of time. Likewise, it is generally not difficult to get students and athletes to practice activities which provide this kind of evaluative intrinsic IF.
Although this kind of trial-and-error learning is made possible because of the intrinsic IF, it is most certainly not an optimal learning environment. The golfer may hit the fairway with his drive on one hole and slice badly on the next. It is difficult for him to learn why the one shot was successful and the other not. Through repeated attempts to reproduce the movement pattern associated with the correct drive he may develop a more consistent, acceptable performance. On the other hand, he may not. For one thing, the learner in this situation gets no IF about his form (topography). For him to systematically attempt to change the topography of his response without outside aid is virtually impossible. He needs the helpful comments of an instructor and/or the visual evaluative IF supplied by a videotape replay of his swing. Coaches and teachers perform a vital role in developing the topography of behavior.
Another problem occurs with young learners. The fact is that good results for a beginner may not always shape a response that will be useful in the long-run. The young basketball player may have to adopt a rather unique form in order to generate enough force to shoot the ball up to a 10-foot-high basket. In a trial-and-error situation the intrinsic IF (making or missing the shot) for this youngster might shape a response that will not be very useful when he gets older and wants to compete at higher levels of skill. In such situations there are two possible solutions. The first would be to adapt the learning environment to the developmental level of the learner so that the IF would shape responses which are more nearly acceptable, that is, lower the basket and use a smaller ball. The second would be to have the learner ignore the IF and concentrate on some other source of IF. The latter solution does not seem promising because results are usually strong secondary reinforcers.
There are also a great many activities which call for long periods of training but do not have much intrinsic evaluative IF. Swimming, track running, and rowing are examples of such activities. In the sheer repetitive training situation, performers need some kind of artificial IF (usually standards of some type) to control and evaluate their performances. The cross-country runner will not improve as quickly as he/she should if he/she is simply told to run 6 miles for practice and is not provided with any IF about the standard of performance.
The problem is even more acute when one considers skill learning rather than work output. Suppose, for example, that a swimming coach attempting to make a slight change in technique, tells the swimmer to push back further at the end of the freestyle arm pull. The swimmer then swims to the other end of the pool executing perhaps 30 responses but receiving no artificial IF about the success of attempts. This is an inefficient use of time, is potentially harmful to the swimmer's performance if done incorrectly, and violates a basic principle of learning. The coach should have devised some method which provided information about the success or failure of each swim stroke. Unfortunately, the situation is rather typical of attempts to change behavior in physical education and sport.
Consider the track runner who is told to run with a lower arm carry and to try that new technique for a certain distance. The athlete then attempts to perform what he/she thinks is the correct interpretation of the coach's instructions. In the meantime, the coach has turned attention to another athlete while the practicing athlete receives no IF about performance of the new technique. This kind of minor change in technique can be accomplished quickly if IF is available at all times during the initial period of change and intermittently thereafter (see Chapter 7).
Learners in a physical education class usually get even less IF due to the large numbers in such classes. Take, for example, the common situation in which the teacher instructs the class as a whole on how to execute a skill such as the backhand drive in tennis. Students are then dispersed over a wide area to practice what they have just "learned." The usual result is that a variety of techniques is displayed by the students and many fail to change in the desired direction during the time spent in class. This is due primarily to the inadequate level of IF available to the learners. Many students adopt strategies that, while affording them some immediate success, prevent them from ever developing the desired level of skill. This happens quite simply because the only IF available to them is the intrinsic KR of whether the shot goes over the net and stays within the boundaries of the court. They tend to adopt strategies that bring about this desired result even if such strategies are not in the direction of the model provided by the instructor. A further problem is created in these situations because once an inappropriate strategy is learned it becomes resistant to change. A great deal of investigation is needed to develop new kinds of IF systems and techniques that will be useful for large group instruction. It is obvious that the teacher or coach cannot develop an efficient learning environment if he is the only source of artificial IF.
An example of the effective use of IF was given by Rushall (1970). The task was to change certain aspects of the technique of a highly skilled swimmer. As has already been mentioned, the difficulty with such a learning task is finding a method by which IF can be made contingent on each response of the learner. Since the intrinsic IF in swimming is not nearly precise enough to bring about the desired changes, some form of artificial IF was necessary. A simple, adequate device for providing IF was a direct-light beam flashlight that was portable and hand-operated. A precurrent training session established the absence of the light as the signal that the new technique was being correctly executed. The presence of the light, on the other hand, indicated an error. The subject in this case was swimming the butterfly stroke. It was possible to stand at one end of the pool and direct the light beam at the swimmer's eyes so that on every breathing phase the swimmer received IF. If three consecutive incorrect responses (stroke cycles) were made, the swimmer stopped for further instruction. In this manner the new techniques were shaped rather quickly. Many repetitions of an undesired response were avoided. Interestingly, this method of providing artificial IF also allowed the swimmer to learn the intrinsic IF cues that were associated with the new techniques. Because of this the swimmer's improved performance continued after the artificial IF was removed.
Another instance can be cited in rowing, where coaches for some time have used megaphones and microphone systems to provide IF to athletes while they are performing. In this case the system was automated one step further. The speed of the boat was monitored by a sensing device attached to it. When the boat achieved a desired speed, a bell was sounded for a short period during each stroke. The boat speed was calculated to be fast enough to win a race and the device was set. The task for the oarsmen was quite simply to keep the bell ringing. Under these conditions the crew rowed faster than they ever had before, even in competition. When the artificial IF was removed, however, the performance returned to previous levels. Evidently the artificial IF had been such that it did not allow for transfer to intrinsic IF sources.
For performance tasks in which the evaluative criteria are fairly simple, automated IF systems will maximize the efficiency of the learning environment. If, for example, force is the desired goal for football lineman in blocking, then accurate, reliable, and immediate IF can be provided by rigging a force plate to a blocking sled. With such a device, practice behaviors could be controlled and new techniques could be learned more quickly.
It is much more difficult to automate IF systems for skills in which the evaluative criteria are more complex. The task of coaching techniques and skills is difficult to evaluate mechanically. The coach is the one who decides whether each repetition of an operant is correct. Even with this limitation, however, the mode of communicating IF should not be difficult to develop. In this case it is the system by which the IF is administered that is important. It must always be kept in mind that changes in behavior are brought about through the consistency and frequency of the IF-reinforcer.
The theoretical differences between the concepts of information feedback and reinforcement are difficult to sort out and these have been treated separately in Chapter 10. What is important to recognize here is that in many cases peculiar to sports and physical education IF is the most effective secondary reinforcer available for the maintenance and development of skilled behaviors. Many learning and training environments could be vastly improved by paying particular attention to the amount, specificity, and frequency of the IF available to the individual performer.
References
Return to Table of Contents for this issue.