THE FEEDBACK FALLACY

One of the greatest disservices perpetrated on society by scientific psychology has been the promotion of the feedback fallacy. The feedback fallacy refers to the notion that evaluative feedback is necessary for improvement of performance and that the more "constructive feedback" a person gets, the quicker performance will improve. A principle which is probably only true if the subjects are blindfolded as they are in the psychological research conducted to support this principle.

Certainly, there is a sense in which the human operates as a servo system, operating to minimize an error signal defined as the discrepancy between the current state and a desired or goal state of the organism. It is not feedback in the sense of this error signal that is referred to here. The feedback referred to here, is external, supplemental feedback provided to the organism by well meaning individuals not directly in the feedback loop.

Rather than the use of feedback to improve performance, I believe that if standards of performance are clear and if the individual is capable of perceiving their performance accurately (they are not blindfolded) then improvement in performance occurs as a result of natural growth processes. For a while I thought that feedback may be necessary for improvement of performance in animals and that the perception of standards was a phenomenon reserved for humans. But then I recalled how many times I've seen students tapping on the side of a Skinner box and gesturing wildly attempting to get a rat to see what it has done wrong. The effect of this "feedback" was to disrupt performance entirely.

In general the feedback fallacy expresses itself in the notion that if you don't tell somebody what they are doing wrong, they won't improve. While there are undoubtedly situations in which feedback of the error signal in this form is necessary, these situations are rare and frequently artificial or contrived. An example of such a situation occurs when you are laying on your back on the floorboards under the steering wheel with your hand up under the dashboard trying to hook something up while someone else is looking into a little vent hole and telling you where your hand is. But in this situation, you are unable to correctly perceive your performance due to situational constraints.

In point of fact, most of the time, when you tell somebody what they did wrong, it produces a relatively strong emotional reaction which tends to interfere with growth and improvement. If you don't believe me, try telling a child that she/he struck out. Yell at them something like "Don't swing at those!!" or "I've told you a thousand times not to swing at those," or any of a thousand other such "feedback" phrases which parents use with children. The effect of such "feedback" is to communicate the parents disappointment in the child, to further erode her/his fragile self image and decrease the likelihood that the child will improve. In fact, such "feedback" is likely to push the child away from the game, making them hate the sport and want to quit.

The child already knew she/he struck out. The emotional energy directed into hate or disdain or disappointment by the "feedback" should have been used for contemplation and interpretation of the experience of the child into a better understanding of the game, the strategies of the pitcher, etc. There is no need for feedback in such a situation. Much more valuable is a "good try" or "it's okay" or "keep trying," which will provide relief from embarrassment or other emotional reactions and enable the child to continue improving. The emotional reactions associated with telling the child what she/he did wrong produce "pulling in" or "withdrawal," "shutting out the experience" rather than the necessary integration.

Assume for a moment, a robot programmed to follow a magnetic stripe painted on the floor of the hallways and offices of a large factory. The robot has sensors near the floor which detect its distance to left or right of the stripe. If the robot is off to the left of the stripe its steering mechanism turns it right until the discrepancy is zero. If it drifts off to the right the steering mechanism turns it left. External feedback is like somebody walking along behind the robot and when it moves a little to the right giving it a shove to the left and when it drifts to the left giving it a shove to the right. The effect of this corrective feedback will be highly disruptive to the performance of the machine.

The detrimental effect of supplemental feedback is particularly evident in skill development situations. Try pointing out each error to someone learning to type, or learning to play the piano. Try yelling "turn left," or "turn right," to someone learning to balance a two wheeler. In this situation performance breaks down completely. The only way the kid will learn to ride the two wheeler is if Dad or Mom runs patiently behind, mainly in silence, for several miles, holding the seat, and waiting for the performance to develop on it's own.

A second aspect of the feedback fallacy is the notion that if positive feedback is given for a sub-standard performance, improvement in performance will be inhibited. This usually expresses itself as a reluctance to reinforce less than the best performance of which the individuals is presumed capable. This too is a fallacy. You can't reinforce a substandard performance. If you holler "Good job" or "Well done" at a child who just struck out, the comment will be taken as sarcasm or lack of reality contact on your part.

AN ALTERNATIVE MODEL OF IMPROVEMENT IN PERFORMANCE

Improvement of performance is actually a function of two perceptual processes. The individual's perception of the standards of performance, and her/his perception of his/her own performance. The computation of the error signal (discrepancy) necessary for the functioning of any servo-system will be done automatically without the need for external, supplementary feedback.

In this model what are the necessary conditions for improvement of performance to occur? They are quite simple. For improvement of performance to occur:

1. the individual must correctly perceive the standards of performance, and
2. the individual must correctly evaluate her/his own performance.

The first of these criteria specifies a need for clearly stated, well articulated, published, agreed on, standards of performance. The second requires an individual who has developed the perceptual skill to correctly evaluate her/his performance. This latter comes primarily through experience and the opportunity to compare performances at a variety of levels of adequacy -- experience and maturity.

Perceptual skills of this type develop slowly, much like an appreciation for good Bordeaux. And once the skill has developed it becomes difficult for the individual to remember what it was like to be unable to distinguish a St. Emillion from a Pauillac. A similar process occurs in learning a game such as football. In the early stages of familiarity, an individual may barely be capable of keeping track of which team is ahead. The subtleties of the game, the strategies employed, escape them entirely. But at some point the mob of players on the field begin to transform themselves into plays with particular names and into identifiable formations, and the strategies of selecting a particular play in a particular situation and the advantages of one formation over the other begin to be apparent. Numerous other examples come from music and art and nearly every area of human experience. In fact, it is just this type of perceptual refinement process or procedure that contemporary cognitive theory identifies as the fundamental psychological unit.

When improvement of performance fails to occur, several possible sources of failure can be identified and only one of them is responsive to additional external feedback.

1. If an individual fails to improve in performance or fails to perform adequately, the problem could be that they are incapable of better performance due to physical or mental limitations. In this situation feedback is particularly undesirable as the individual is left feeling that they have failed and there is nothing they can do about it.
2. When an individual fails to perform adequately, the problem could be that they are not correctly evaluating their own performance. That is they may understand the standards perfectly well, and agree with them, but think that what they are doing meets the standards. This is a rare condition which generally only occurs when the individual is out of touch with reality, lacks maturity, or has a deficit in their perceptual processes.
1. If the individual is out of touch, it is unlikely that feedback will help.
2. Maturity is little affected by feedback as well.
3. Individuals with deficits in their perceptual mechanisms are the exception to these comments. Three types of deficit can be identified.
1. Individuals with deficits in perception can be responsive the supplemental feedback and, as they are most likely aware of these deficits, are normally most appreciative of the assistance. (Witness the fellow on the floorboards with his hand up under the dashboard.) This latter is the situation usually studied in the laboratory. In laboratory experiments on feedback, the individual's perceptual mechanisms are generally handicapped (by a blindfold) and the deficit replaced in one group by supplemental feedback.
2. A second type of perceptual deficit comes about when the person has been practicing errors; when they have been doing it wrong for so long that it feels right. Errors in the way one holds or swings a golf club or tennis racquet, errors in the way one delivers a bowling ball, and bad postural habits are examples of this that come to mind. But this situation also occurs when an individual has learned incorrect or inappropriate standards of behavior in any area of life. Frequently the person doesn't know that their behavior is incorrect or inappropriate. It takes the patient training and modeling of an experienced coach to change this type of ingrained error.
3. This situation also arises when due to the natural perceptual limits of the individual, he or she is not able to detect performance discrepancies. In this situation additional information, frequently coming from tools or instruments (a microscope or magnifying glass for example), is most beneficial. However, what the person gets in this situation is additional data. Not evaluative feedback. In general it is beneficial to provide an individual with new data or interpretations or comparisons of the data which were not available to her/him. But what makes this effective is the fact that the individual is obtaining previously unavailable data, not that it is evaluative feedback.
3. Failure to improve could result from a failure to correctly perceive or agree on the performance standards.
1. If the standards are unclear, or unstated, mis-perception of the standards is a distinct possibility. In this situation, the individual doesn't need feedback concerning her/his performance. They need to be told what the standards are.
2. The standards may be clear and the individual fail to understand them. Once again, the individual needs training on the standards, not feedback concerning her/his performance. Although in this case, the performance of the individual can provide the trainer with a rich source of examples which are relevant and close to home, to be used in explaining the standards. Frequently, skilled teachers use this technique, in just this situation, with great success and it is mistaken for giving the individual feedback on her/his performance. However, what makes this effective is that the individual is learning the standards, not that they are receiving feedback.
3. The individual may be using a different set of standards than the evaluator. Most of the problems of incorrigible employees who won't improve, fit this category. Because of different orientations, different paradigms, different frames of reference, different political or religious belief systems, two individuals can have dramatically different standards of performance. When this happens, feedback is frequently completely ineffective.
1. A good example of this occurs when a student who has been taught to do anything necessary to get ahead and who believes that everybody else cheats on tests and exams in college, is caught for cheating and subsequently punished. The punishment doesn't change her/his attitude toward cheating, rather the student is frequently left with a sense of bewilderment, a lack of understanding of what happened. The dissonance is frequently resolved by deciding that she/he is being picked on or that the instructor is crazy.
2. Sometimes this situation occurs when the individual knows and understands the standards of the evaluator but disagrees with those standards. In this situation the individual chooses not to comply. In fact if the individual were evaluated by another standard, their performance may be improving substantially. In this situation as well, feedback is useless. Feedback that the performance is inadequate simply confirms the commitment of the evaluator to standards which the individual being evaluated considers inappropriate. This is particularly a problem in organizations in which heterogeneous groups of individuals are thrown together in some sort of association. Large Psychology Departments are a good example. The variety of paradigms from which members of a psychology department can come is so broad that it is little wonder that conflict frequently exists.

One of the primary advantages of this model is the diagnostic specificity which it provides. What, according to this model, can be done to improve performance in individuals who have been unresponsive to feedback for some time. First, it must be determined that improvement is within the capability of the individual. Then efforts toward improvement should be focused on clarifying and agreeing on standards. Clarifying standards is a communication problem. Agreement can take a great variety of forms including agreeing on a plurality of standards.

In any case, according to this model, once the standards and the performance are correctly understood and agreed on, performance will improve as a result of natural growth processes and feedback is not generally necessary. In fact it is generally disruptive. Providing evaluative feedback in a growth process is much like telling a child to be taller.

Copyright © 1996 Steve Falkenberg