Ghost in the Box, Chapter 1 – Models of Biofeedback Training


Models of

Biofeedback Training

The model or set of models a psychologist believes in determines to a great extent, the kind of research he does, and the type of explanations he develops.

Introduction to Experimental Psychology, 
Matheson, Bruce, and Beauchamp, 1974

Science advances by conceptualization and experimentation. Inappropriate concepts and models result in poor experimental design, unsuccessful research, and years of wasted time and effort. Appropriate conceptualizations and models can result in the rapid development of an emerging science.

Our first task is to determine an appropriate model for biofeedback training. Our second task is to examine current conceptualizations and designs that underlie biofeedback research. Our thesis is that, with a few notable exceptions, biofeedback research has lacked clear and appropriate conceptualizations and has lacked appropriate experimental design.


Conceptualizations of Biofeedback

An Appropriate Model

What is an appropriate model for biofeedback training? Is biofeedback like a drug? Is it like training a rat? Is it like training an athlete to excel in a sport or teaching a student skills, or is biofeedback something else altogether? The answer depends on

Shellenberger and Green 2

how biofeedback training and the goals of biofeedback training are conceptualized.

In the early days of biofeedback, excitement and interest were sparked by a clear goal: biofeedback training would be a tool with which individuals could gain voluntary self regulation of various psychophysiological processes. The interest in self regulation was expressed in titles chosen for the publications, Biofeedback and Self-Control (Aldine Books) and Biofeedback and Self-Regulation, the Journal of the Biofeedback Society of America. Self regulation was fascinating, since the autonomic nervous system and aspects of the somatic are considered “involuntary.” In addition, self regulation was potentially of great benefit for symptom reduction in the treatment of psychosomatic disease. It was exciting to find that by using information feedback, trainees could learn to regulate psychophysiological processes without drugs.

What is the nature of this tool called biofeedback training? Biofeedback training is the use of instruments to feed back psychophysiological information to a person. The information is referred to as ‘psychophysiological” because psychological processes are reflected in physiological functioning. Biofeedback instruments are designed to monitor, amplify and feed back a variety of biological processes such as heart rate, blood pressure, muscle tension, blood flow in the hands and feet, and brain waves. The trainee uses the information from her/his body to learn to make changes in the psychophysiological process being monitored. The information is helpful in learning to regulate mind and body in the same way that information from the dart board is helpful in learning to play darts. The trainee uses the information to become conscious of, and voluntarily create, the physiological and psychological states that produce the desired physiological change. The key here is consciousness because consciousness is necessary for psychophysiological self regulation. And the key to consciousness is feedback of information.

Two analogies have been used to describe the function of biofeedback instrumentationOne analogy describes the instrumentation as removing a blindfold”This use of instrumentation is analogous to removing the blindfold from the novice who is trying to shoot a basket . . . If our novice basketball player were blindfolded so that he did not have the feedback of seeing whether he was suc-

Ghost in the Box 3

ceeding or failing, he could not learn” (Miller, 1975, p. 367). Without information feedback, learning cannot occur. This is true in sports, education, social learning, and certainly in biofeedback training in which the trainee is attempting to gain voluntary control of a normally involuntary, ”blindfolded,” and unconscious process.

mirror is the other appropriate and often used analogy to describe the function of the biofeedback instrument. Like a mirror, the instrument enhances consciousness and psychophysiological self regulation by reflecting the psychophysiological process being regulated. Like a mirror the only requirement of the instrument is that it provides a true reflection, i.e. useable and accurate information. Like a mirror, the usefulness of the biofeedback instrument is dependent upon how the information from it is used. Like a mirror, the biofeedback instrument has no inherent power to create change. The human using the biofeedback mirror has the power to control the process being reflected.

When the mirror characteristics of the biofeedback instrument are understood, the rest of the model unfolds: the trainee uses the information to gain self regulation of these processes, and learning progresses either through trial-and-error, or learning progresses through systematic training.

Biofeedback training is the process of mastering psychophysiological self regulation skills, with the aid of information from a biofeedback instrument, and is similar to skills learning in any activity such as sports, music or education. This is the model that we develop in this manuscript: we refer to it as the ”mastery model.”

The essential ingredients Of biofecdback training are those of training in any complex skill: clear goals, rewards for approximating the goals, enough time and practice for learning, proper instructions, a variety of training techniques, and feedback of information. The essential ”ingredients” of the user of the information from the biofeedback machine are those of any learner of a complex skill: consciousness, cognitive understanding, language, positive expectations, motivation, and positive interaction with the coach, teacher, or therapist. The implications of this model for biofeedback training, in clinical practice, and in research are vast.

To a large extent, biofeedback as a mirror, and biofeedback train-

Shellenberger and Green 4

ing as mastery of self regulation skills, are concepts that have been accepted by biofeedback clinicians. With a few notable exceptions, occurring early in the history of biofeedback research and recently, these concepts and their implications have not been accepted by researchers.


Biofeedback Research Models

Iour review of over 300 theoretical papers and research reports on biofeedback training, we find that the most common conceptualizations of biofeedback training are derived from two models: (1) operant conditioning with laboratory animals and (2) specific effects of drugs.

The Operant Conditioning Model.

Many of the early biofeedback studies were done by researchers trained in operant conditioning methodology with laboratory animals, such as Benson, Kimmel, Miller, Shapiro, Sterman, Taub, and Tursky. These researchers applied the methods, language, and goals of animal research to biofeedback training with humans. The procedures are described as: ‘sensorimotor EEG operant conditioning” (Sterman, 1977); “operant electrodermal modification” (Shapiro & Watanabe, 1972); “learned control of human cardiovascular integration through operant conditioning” (Schwartz, Shapiro, & Tursky, 1971); “classical and operant conditioning in the enhancement of biofeedback” (Furedy & Riley, 1982); “operant conditioning of heart rate slowing” (Engel & Hansen, 1971).

Using the operant conditioning model of learning, researchers focus on concepts such as the stimulus“the goal is to bring [biofeedback subjects] under some degree of stimulus control” (Alexander, 1975, p. 216); the reinforcer“among the strategies suggested [for the study of the active treatment effect of biofeedback were delaying reinforcement, varying reinforcement schedules…” (Katkin & Goldband, 1979, p.186); “You can see that feedback did help subjects maintain a higher skin temperature than subjects who did not have feedback . . . It looks

Ghost in the Box 5

like biofeedback acts like a reinforcer here and that subjects who do not have it seem to be extinguishing more rapidly than subjects who do” (Surwit, 1982, p.240); and the contingency between stimulus and response: With biofeedback, however, I think we have been somehow uniquely blessed. No matter whether one conceives of the process of biofeedback training as operant conditioning, motor skills learning, or self control in a purely phenomenological sense, it is the contingency between biological behavior and the feedback stimulus that all theories predict is responsible for determining change” (Hatch, 1982, p. 381). And biofeedback is effective “if and only if it is shown that it is the contingency between the target behavior (HRD) [heart rate deceleration] and the reinforcement (i.e., information or feedback about HRD) that is responsible for the increase in HRD” (Furedy, 1979, p. 83).

Furthermore, some researchers assume direct parallels between human learning and laboratory animal learning. To explain the methods and poor results of their study with Raynaud’s patients, Guglielmi Roberts and Patterson (1982) write:

A third argument that could be used to challenge the results of this study might be that the subjects’ ignorance of the feedback-relevant response inherent in double-blind studies necessarily led to inadequate training and therefore to negative findings. The success in training animals, obviously unaware of the target response, to control a variety of autonomic functions would seem to invalidate this contention (Guglielmi et al., 1982, p. 117)

The argument is that since laboratory animals are “unconscious of the target response” then obviously humans need not be aware either. The conditions that apply to rats also apply to humans.

The Drug Model

“The removal of the contingency between biofeedback and behavior is logically similar to removing the active ingredient from a medicine” (Hatch, Klatt, Fitzgerald, Jashewy, & Fisher, 1983,

Shellenberger and Green 6

p. 411). “In this regard, pharmacological medicine provides an appropriate model for treatment evaluation.” (Furedy, 1985, p. 159). Application of the drug model to biofeedback training results in conceptual issues and research designs unique to the model that have occurred repeatedly in biofeedback research. It is assumed that like a drug, biofeedback has “specific effects” and that these specific effects must be demonstrated independently of any “non-specific” effects. In addition, the drug model insists that the placebo effect is a hazard to the study of the specific effects of biofeedback, and must be controlled for or eliminated (Beatty, 1982; Furedy, 1985; Katkin & Goldband, 1979; Miller, 1976, Price, 1979).

To study these specific drug-like effects, designs appropriate to drug research are adopted for the study of biofeedback, primarily the double-blind design. (Hatch et al., 1983; Guglielmi, Roberts & Patterson, 1982). “The function of the double-blind arrangement is that it separates placebo effects from specific effects” (Furedy 1985, p. 159).


The Official Doctrine

A combination of concepts taken from drug effects and animal research models has prevailed in biofeedback research, so much so that we call it the “official doctrine.” The “official doctrine” approach has been the scientific model in biofeedback research, and is still rewarded by research funding, publications, and university degrees. But are the concepts of the official doctrine derived from drug and animal models applicable to biofeedback training? The answer to this question is crucial.

To assess the applicability of the official doctrine to biofeedback training the models must be examined conceptually and empirically. Empircallywe must examine the research methodology, results and conclusions that follow from the models of the official doctrine. Conceptuallywe must examine whether or not the conceptualizations, the language, and the goals of the models accurately describe the phenomena being studied. The latter task is facilitated by employing a concept that has evolved from the history of science, the category mistake.

Ghost in the Box 7

Category Mistake


The concept of “category mistake” is useful in analyzing the conceptualizations that underlie biofeedback training (Hesse, 1970; Ryle, 1949; Ryle, 1965). A category mistake occurs when conceptualizations appropriate to one category are inaccurately applied to another. Anton Mesmer made a category mistake in his explanation of hypnosis. This is illustrated in the investigation of his powers by a group of distinguished scientists attempting to determine the veracity of Mesmer’s claim that he had access to a powerful healing force. The scientists set up an experiment in which a sick boy was brought to Mesmer to heal. Mesmer said that the healing power could be put into a tree and when the boy touched the tree he would be healed. The boy was told that he would be healed, and he was healed, but he touched the wrong tree. The scientists concluded that Mesmer was a fake. Mesmer’s category mistake was to apply a concept from the category of matter, magnetism, to the properties of an interactive process between human beings. The scientists also made a category mistake by concluding that Mesmer was a fake, thus dismissing an important phenomenon. As a result of these category mistakes, hypnotherapy was prevented from reaching its full fruition for many years (Frankel, 1976).

In the initial investigation of new phenomena, category mistakes are likely because familiar conceptualizations appropriate to known phenomena are inaccurately applied. The category mistakes are often subtle and difficult to recognize because they appear to be correct. As attempts have been made to develop and understand biofeedback training through research, category mistakes have been made that arise from faulty conceptualizations.


Category Mistakes and the Ghost in the Box

The essence of the official doctrine derived from drug and animal models is that the biofeedback instrument, or “biofeedback” when this means merely the use of a machine, has an inherent power with specific effects. These effects are either like a drug, as in the

Shellenberger and Green 8

drug model, or like the stimulus, reinforcer or contingency, as in the operant conditioning model. “In summary, stimuli have innate and acquired properties which enable them to control behavior” (Engel, 1979, p. 171). To conceptualize biofeedback in this way is to make a category mistake. The biofeedback instrument is merely a mirror that provides information. It has no inherent power to create change or control behavior. It would be a category mistake, indeed superstitious, to believe that a mirror has power, and it is a category mistake and superstitious to believe that biofeedback has a special “specific” power to create change. This belief is superstitious because it assumes causality where none exists. It was superstitious to believe that Haley’s Comet caused the death of the king because the king died when the comet was overhead, and it is superstitious to believe that the biofeedback instrument, or its signal characteristics, cause behavior change when the human creating the change is using a biofeedback instrument.

A characteristic of information is that it has no power independently of the userthe user’s goalsand the environment of the user. An uninterested general psychology student who is required to participate in a biofeedback experiment may find the biofeedback information boring. A tension headache patient, however, may find EMG information rewarding because her goal is to reduce her pain. But it would be a category mistake to believe that the “biofeedback mirror” has an inherent and specific effect that brings behavior under its control.

A second category mistake arises from applying a simplistic model of operant conditioning to biofeedback training. We noted in the discussion of the operant conditioning model that researchers have attributed behavior change through biofeedback training to a variety of variables~the stimulus, the reinforcer, the contingency between stimulus and response, or the contingency between the response and the reinforcer. To suggest that behavior change through biofeedback training is determined or controlled by these variables, is to borrow concepts appropriate to one category, operant conditioning with laboratory animals, and apply it to another category, human learning. This is a category mistake. The category “rodent” is different from the category “human” and the category “animal learning” is ipso facto dif

Ghost in the Box 9

ferent from the category “human learning.”   Consequently laboratory research with animals is categorically different from research with humans. Unlike rats, humans are not microcephalic, they have self awareness, and they bring to research complex and uncontrolled variables: (1) environment, e.g. job and family stressors, insurance reimbursements, and workman’s compensation; (2) an elaborate language; (3) beliefs and expectations; (4) motivation; (5) and a host of psychophysiological and somatic illnesses for which they are seeking treatment. This enumeration of human characteristics as evidence of the categorical difference between humans and rodents may seem obvious. Nevertheless, in ”official doctrine” research these differences are not taken into account or are purposefully ignored.

Working with rats and pigeons, experimenters have derived principles of learning that appear to govern behavior. Indeed in animal laboratory settings operant conditioning theory may apply, and perhaps its application to humans is appropriate in understanding and treating particular cases, such as the mentally retarded, infants, phobic, and impulsive patients, And it is certain that reinforcement motivates behavior acquisition. However, while learning to peck at a green dot and not a red one, or learning to run a complicated maze, or jump through a hoop, may represent the absolute pinnacle of animal learning, responses conditioned to external stimuli represent the most rudimentary form of human learning.   When operant conditioning theory is applied to human learning in biofeedback training, the most rudimentary form of learning is studied through the methodology of the model. The theories, goals and methods appropriate to the pinnacle of human learning-higher mathematics, virtuoso musical performance, gold medal ice skating, and self regulation of a complex psychophysiological process—are ignored. Bar-pressing behavior becomes the model for human learning. It will be noted that, just as biofeedback can be formulated in terms of operant conditioning or in terms of voluntary control, so the lever-pressing example could have been formulated as either operant lever-press conditioning with light as reinforcement or as the rats’ voluntarily ”emitting” lever presses “in order to get” the light” (Furdey & Riley, 1982, p.84).

Because these models conceptualize “biofeedback” as having a specific drug-like effect, or innate or acquired properties like

Shellenberger and Green 10

stimuli and reinforcers, and assume an inherent power that does not exist, we call them the ”ghost in the box” approaches to biofeedback training.

In summary, our thesis is that the official doctrine incorporates two major category mistakes:

Category Mistake #1: equating the properties of objects such as drugs with information processing by humans of signals from the biofeedback instrument;

Category Mistake #2: equating animal learning with human learning, the latter involving goals, cognitive and emotional processes, language, and complex uncontrolled environmental variables.


Consequences of the Official Doctrine and

Category Mistakes


To determine the appropriateness of the official doctrine models empirically, we must examine the research methodology that is derived from the models. We propose that at least twelve conceptual and methodological errors arise from the category mistakes of the official doctrine. These errors are not mutually exclusive. They are interrelated and produce similar research results and conclusions, because they are derived from common category mistakes. The interrelationship of these errors and the need to analyze them carefully and precisely leads to recurrent themes in our discussion. We believe that this is useful in helping to clarify the issues and problems in biofeedback research. These errors contribute to unsuccessful biofeedback training and have hindered the development of the field and its acceptance in the scientific and medical communities.

In Chapter Two, we examine the twelve conceptual and methodological errors and the research results and faulty generalizations derived from them.

We then briefly examine the conceptualizations and methods that underlie successful biofeedback training. A training model appropriate to the goal of maximizing psychophysiological self regulation and symptom reduction is proposed in Chapter Two, and developed in detail in Chapter Four. We refer to this as the 

Ghost in the Box 11

“mastery model.” The model incorporates concepts and methods of high performance training of athletes. As in athletic training, mastery is the sine qua non for successful biofeedback training.


[End of Chapter 1]


Attention: The Ghost in the Box is “Shareware”!
Please Register Your Copy Today

This internet publication of the historic monograph “From the Ghost in the Box to Successful Biofeedback Training” is itself an historic event — the first known publication of a masterpiece previously-published document as “shareware”. 

For years computer software has been published under this “honor system”.  You can download a program for free, try it out, and see if you like it. And, if you continue to use the program, you are honor-bound to send the modest registration fee to the author. 

If you download and read this Internet Edition of GHOST, either the whole book or any one or more of its chapters, and if you allow its message to influence your thinking about Biofeedback, you are asked to submit the modest sum of FIVE DOLLARS directly to the book’s authors, Bob and Judy.  That’s even a bargain, since the original 1986 publication, which is herewith reproduced in its entirety, cost $9.95!  Unfortunately, it has been out-of-print for several years, but would surely cost more if reprinted today.

License.  Payment of the $5.00 registration fee entitles the reader to print one (1) copy of the entire text, or any part of the entire text, for personal use.  Up to ten additional copies may be printed, provided that this notice is always included (once) and each recipient understands that the shareware fee applies to each and every printed copy of the book or any chapter of the book.   Reproduction beyond the scope of this license is a violation of US and International Copyright Laws.

To Register your copy of “Ghost”, print this page and send it with your check or US$ 5.00 cash to:

Bob Shellenberger & Judy Green
c/o Psychology Department
Aims Community College
PO Box 69
Greeley, CO 80632 USA

Dear Bob and Judy:

Thanks for making “The Ghost in the Box” available again. 

Name: _________________________________________

Address: _______________________________________

City/State: ______________________________________

Email Address: __________________________________

Remember, send only one registration per person, regardless of how many chapters you have. 

A revised edition of Ghost is in the planning stages; if you do include your name and address, you will be notified if and when it becomes available. 

A brief description of your professional involvement in biofeedback would be of great interest to the authors. 

Comments, suggestions and criticism are most welcome.  Please feel free to make your suggestions or comments here.  



Leave a Reply

Your email address will not be published. Required fields are marked *