Download On Some Research-Community Contributions to the Myth and

293
Back to Biofeedback...
Home Page
International Journal of Psychology: 4 (1987) 293-297
Myth_Symb87.doc
ON SOME RESEARCH-COMMUNITY CONTRIBUTIONS TO THE MYTH AND SYMBOL
OF BIOFEEDBACK
JOHN J. FUREDY
Department of Psychology, University of Toronto, Toronto, Out. (Canada)
(Accepted 25 September 1986)
Key words: Biofeedback: Instrumental conditioning: Non-contingent control: Feedback vs feedforward; Definitional
issue
This paper suggests that some of the responsibility for the misrepresentations of biofeedback noted
by Kimmel (Int. J. Psychophysiol.. 19X6, 3: 211-21X) rests with the research community's treatment
of the topic. With regard to the "myth" of biofeedback, it is argued that, after Miller's (Science, 1969,
163: 434-445) influential paper, most researchers' treatment of the proper-control-for-biofeedback
problem was one that selectively lowered normal methodological standards. and therefore produced
results that were improperly interpreted as evidence for biofeedback (or instrumental conditioning). The
shift from instrumental conditioning to biofeedback terminology, which was based primarily on political
ideology rather than on logic, may have been made easier by a reluctance on the part of the scientific
community to engage in reflective analyses of concepts and definitional problems.
INTRODUCTION
Kimmel's recent discussion of the myth and
symbol of biofeedback is both timely and, as he
suggests, appropriate for scientists "who have
played important historical roles" (Kimmel, 1986,
p. 298). The basic points he makes, the broad
distinctions he draws, and the condemnations he
issues are sound and necessary in understanding
an attractive, yet difficult concept. The biofeedback concept, as I have argued elsewhere (Furedy,
1984b; 1985), has been all too frequently applied
in a superstitious or 'snake-oil', rather than scientific fashion to behavioral problems.
However, there are some more subtle factors
that have contributed to the myth and symbol of
biofeedback. In this addendum to Kimmel's remarks, I suggest that the treatment of the biofeedback phenomenon by the research community
Correspondence: J.J. Furedy, Department of
Psychology, University of Toronto, Toronto. Ont.,
Canada, M5S 1Al.
0167-X760/X7/$03.50 © 1987 Elsevier Science
Publishers B.V. (Biomedical Division)
contributed to the resultant conceptual confusion.
Specifically, I shall argue: that most researchers'
treatment of the proper-control problem contributed to the myth of biofeedback; and (more
speculatively) that there was a shift from operantconditioning to biofeedback terminology among
researchers that was not driven by purely scientific
or logical considerations, and that therefore contributed to the (political) symbolism that grew
around the biofeedback concept.
MYTH
I agree that it is grossly distortive to claim that
biofeedback "was the product of anti-establishment thinking, anti-scientific methodology, anticonventionality" (Kimmel, 1986, p. 214). However, myths generally have a grain of truth that
accounts for at least part of the belief in the myth
itself. In the case of biofeedback or instrumental
autonomic conditioning, that grain of truth is the
294
fact that once the biofeedback phenomenon had
gained popularity among researchers (and that
occurred with Miller's (1969) influential Science
publication), the scientific community relaxed its
methodological standards with respect to the appropriate control conditions for accepting a result
as indicating the effects of biofeedback.
The community, including Miller (1969), agreed
that biofeedback was a form of instrumental conditioning. The issue of whether a given phenomenon constituted instrumental conditioning was
familiar to anyone conversant with the animallearning literature of the late fifties. At that time,
learning theorists wished to determine whether
non-biological reinforcers such as weak lights were
really reinforcing for, or produced instrumental
conditioning in. rats (e.g. Berlyne, 1960). The first
proponents of this view thought it sufficient simply to show that the rate of lever pressing ('instrumental conditioning') in rats would increase if
such a light followed each lever press. However,
the critics soon pointed out that such an increase
was not sufficient for concluding that instrumental conditioning had, in fact, occurred, because
there was no control for the possibility that certain properties of the light qua stimulus were
responsible. For example, the light may have energized the animals, raised their overall responsiveness, and hence raised their lever-pressing rate.
On the other hand, if it is assumed that overall
responsiveness competes with (and therefore reduces) the lever-pressing rate, the light may have
calmed the animals, lowered their overall responsiveness, and in this way increased their lever
pressing. Accordingly, without some control for
the non-contingent effects of the light, no researcher accepted the simple result of light increasing lever-pressing as evidence for instrumental conditioning-only the unsophisticated would
accept the 'myth' on such an insubstantial evidential basis.
The form of non-contingent control that was
developed in the light-reinforcement studies was
the 'yoked control'. In this arrangement, the 'master' animal's lever-press performance would control the light occurrence both for itself and another
animal 'yoked' to it. Then, any superiority in the
master animal relative to the yoked control animal
could more reasonably be attributed to the performance-reinforcement contingency, thereby allowing the inference that the phenomenon of instrumental conditioning had occurred with this nonbiological reinforcer. It is of interest to note that
Kimmel's early work on instrumental autonomic
(GSR) conditioning (e.g. Kimmel and Hill, 1960)
was extremely rigorous, and provided such a
yoked-control condition at a time when the research community was much more sceptical than
credulous concerning the possibility that autonomic responses could be instrumentally conditioned.
On the other hand, both in the period immediately following Miller's (1969) influential
paper, and until the present, the most frequent
control condition used by researchers has been the
no-feedback control. Yet, this is formally equivalent to the methodologically inadequate no-light
control that was used in the initial light-reinforcement studies, but quickly abandoned for the
yoked-control design. Moreover, with human subjects, the no-feedback control is particularly naive,
because of the presence of placebo effects that are
both considerable, and vary markedly as a function of time and place.
It is true that the so-called 'bi-directional' control used in the studies reported by Miller (1969)
is superior to the no-feedback control. The bidirectional control may be illustrated by taking
heart rate (HR) change as the target response
(TR). In that case, the performance of subjects
reinforced for HR acceleration would be
contrasted with that of subjects reinforced for HR
deceleration. However, the acclaim accorded to
the bi-directional control appears inappropriate.
Formally, the bi-directional control is analogous
to testing the light-reinforcement effect by having
the control group reinforced for behavior that is
incompatible with lever-press behavior (just as HR
acceleration
is
incompatible
with
HR
deceleration). This is because, rather than varying
the critical factor (contingency), the bi-directional
control varies the nature of the TR. More
specifically, unless the rate of reinforcement is
equated between the acceleration and deceleration
conditions, any obtained difference between the
two conditions may not reflect the biofeedback
phenomenon, but merely a
295
difference between the two that is unrelated to
contingent reinforcement. It is important to recognize that the rate of reinforcement has not been
equated between conditions in bi-directional control experiments, except for a handful of studies
(e.g. Riley and Furedy, 1981; Schober and Lacroix,
1986). It is at least arguable, therefore, that even
the bi-directional control biofeedback studies are,
to an unknown extent, representative of mythical
rather than real biofeedback effects.
The credulous rather than sceptical approach to
the control problem by most researchers following
the Miller (1969) paper is particularly notable,
because many of these researchers came to biofeedback from a learning-theory background. It is
instructive to compare the level of sophistication
shown by biofeedback researchers to that shown
by those concerned with the earlier problem of
whether weak light produced instrumental conditioning. Soon after the yoked-control design was
employed, an influential paper by Church (1964)
strongly criticised this method of assessing the
contingency effect. However, this paper is not, as
is often supposed, an argument against the noncontingent control: it is only an argument for a
proper non-contingent control. In my view, the
problems that Church raises can be mitigated
either by ensuring that individual differences are
minimal, or by equating the reinforcement schedules not only between, but also within groups (e.g.
Riley and Furedy, 1981). There are also other
problems that are specific to biofeedback. Of these,
two sets are ethical problems involved with not
providing one group with feedback, and methodological problems involved with the possibility of
subjects in the false-feedback condition discovering their status, and thereby losing motivation. As
detailed elsewhere (Furedy and Riley, 1982; Furedy, 1985) these two sets of problems can be
solved, respectively, by using within-subject
designs and a degraded rather than a false-feedback condition.
However, arguments over how to provide adequate non-contingent control conditions already
presuppose that, in the scientific assessment of
any phenomenon, it is specific rather than placebo
effects that have to be sought. It is, perhaps, not
surprising that the specific-effects approach had to
be defended as a minority position, to a predominantly clinical rather than basic-research audience
(Furedy, 1985). However, some years before at a
conference dominated by workers whose main emphasis was basic research, I found that the
specific-effects, non-contingent control approach
advocated by Furedy and Riley (1982) also represented a minority, though not a singular (see e.g.
Orne, 1982) view. The extent of the unpopularity
of the specific-effects approach was emphasized
when the alternative, placebo-oriented, 'biofeedback-package' idea was defended in print both by
the discussant of our paper (Mulholland, 1982; for
reply, see Riley and Furedy, 1982), and by the
reviewer of the book (Tursky, 1982) in which the
paper appeared (Lehrer, 1984a, 1984b; for replies,
see Furedy, 1984a, 1984b). The specific-effectsoriented presupposition that was shared by researchers concerned with the light-reinforcement
effect in rat lever-pressing does not seem predominant in the biofeedback research community of
the eighties. And while the particular ways of
achieving the non-contingent control are matters
of opinion, it is indisputable that the more general
approach that fails to adopt a specific-effectsorientation to biofeedback represents a lowering
of methodological standards, and thereby tends to
propagate 'guru' myths I. Accordingly, while the
'revolutionary' myths that some 'gurus' have promoted are grossly false, and are quite properly
attacked by Kimmel (1986), it seems to me that
the loose treatment of the biofeedback phenomenon is partly due to the failure of the research
community, after Miller's (1969) paper, to apply
the same standards of methodological rigour to
the biofeedback phenomenon that are applied to
other instrumental-conditioning phenomena, and
that were applied to Kimmel's own earlier work
(e.g. Kimmel and Hill, 1960).
In traditional systems-theory terms (see e.g. Anliker,
1977; Tursky, 1982), the confusion is between feedback
and feed forward effects. The latter are instructional
(in humans) effects that produce control that is not "the
result of the response-contingent signal ('feedback')
provided by the experimenter". (Riley and Furedy,
1982, p. 137).
296
SYMBOL
I agree with Kimmel (1986) that to link the
biofeedback/conditioning distinction with that
between liberty and fascism is to take a position
that may have some political and rhetorical validity, but one that is bereft of any logical or evidential merit. And his anecdote of his lecturing experience provides a vivid illustration of the havoc
that sophistry can wreak on the logic of using a
term like the politically loaded one of 'biofeedback'.
Nevertheless, the 'symbol of liberation' in moving from ' instrumental autonomic conditioning' to
' biofeedback' does not seem to have been actively
questioned by the writings of the scientific community. Some practioners may wish to dismiss
such terminological problems as 'merely semantic',
but traditionally it is the academic community that
is concerned with, and reminds others about,
the need for defining one's terms, and for drawing
distinctions on the basis of logical, rather than
rhetorical considerations. Indeed, it can be argued
that this 'Socratic' (Furedy and Furedy, 1982;
Kimble, 1984), reflective feature is what distinguishes epistemic from other human en deavours. The research literature on biofeedback,
with a few notable exceptions (e.g. the analysis by
Kimmel [1978] of the concept of voluntariness as
it relates to biofeedback), contains little by way of
such reflective articles that are concerned with
analysing the concepts involved, rather than reporting on the successes and failures of various
'biofeedback' preparations. More generally, North
American research psychology (which is the basis
of the biofeedback movement) has shown a lack of
concern with so-called 'philosophical' issues, and
with arguments over the definitions of scientific
terms. For example, after Ax (1964) and Stern
(1964) raised the question of how to define
psychophysiology, the question was never again
raised in the journal in which their articles were
published. Yet it cannot be that they had solved the
definitional problem, if only because their views
were incompatible, so that both could not be
right. Rather, it appears that the scientific community did not consider the problem worthy of
further attention, and therefore shelved it in favour
of the seemingly more practical task of 'getting
on' with their empirical research.
This is not to argue for a return to the days
when psychologists acted like philosophers, and
did nothing else but reflect about their subject. It
is only to suggest that if scientists are not prepared to spend some of their time reflecting about
their subject and discussing conceptual and definitional issues, they should not be surprised if others
with a more practical bent employ terms like
'biofeedback' for their own symbolic, as well as
mythical, use.
ACKNOWLEDGEMENTS
This paper was prepared with partial support
from grants from the National Science and Engineering Council of Canada, and a Sabbatical
Leave Fellowship (1985-6) from the Social Science
Research Council of Canada. I am indebted to
Hal Scher, Christine Furedy, Diane Riley, Donna
Shulhan, and Amanda Walley for comments on
an earlier draft.
REFERENCES
Ax,
A.F.
(1964)
Goals
and
methods
of
psychophysiology. Psychophysiology, 1: 8-25.
Anliker, J. (1977) Biofeedback form the perspective
of cybernetics and systems science. In J. Beatty and
H. Legewie (Eds.), Biofeedback and behavior.
Plenum, New York.
Berlyne, D.E. (1960) Conflict, arousal and curiosity.
McGraw-Hill. New York.
Church, R.M. (1964) Systematic effects of random error
in the yoked control design. Psychol. Bull.. 62: 122l37.
Furedy, J.J. (1984a) Booers. beware biofeedback
boosting. Contemp. Psychol., 29: 524
Furedy, J.J (1984b) Specific versus placebo effects in
scientific versus snake-oil medicine. Contemp.
Psychol., 29: 599.
Furedy, J.J. (1985) Specific vs. placebo effects in
biofeedback: science-based vs snake-oil behavioral
medicine. Clinical Biofeedback and Health an
International Journal, Vol. 8. pp. 110-118.
Furedy, J.J. and Furedy. C (1982) Socratic versus
Sophistic strains in the teaching of undergraduate
psychology: implicit conflicts made explicit. Teach.
Psychol., 9: 14-20.
Furedy, J.J. and Riley, D.M. (1982) Classical and
operant conditioning in the enhancement of
biofeedback: specifics and speculations. In L. White
and B. Tursky (Eds.), Clinical biofeedback Efficacy
and Mechanisms. Guilford. U.K.
297
Kimble, G.A. (19X4) Psychology's two cultures. Am.
Psychol., 8: 833-839.
Kimmel. H.D. (197X) Making involuntary behavior
voluntary: what does this do to the distinction?
South. J. Philos., 16: 213-226.
Kimmel, H.D. (1986) The myth and symbol of
biofeedback. Int. .J. Psychophysiol. 3: 211-218.
Kimmel, H.D. and Hill, F.A. (1960) Operant
conditioning of the GSR. Psychol. Rep., 7: 555-562.
Lehrer. P. (1984a) Yes. a cautious (and limited)
boost for biofeedback. Contemp. Psychol., 29: 524.
Lehrer, P. (1984b) Biofeedback: Not penicillin but not
snake oil. Contemp. Psychol., 29: 599.
Miller, N.E. (1969) Learning of visceral and
glandular responses. Science, 163: 434-445.
Mulholland, T.B. (1982) Comments on the chapter by
Furedy and Riley. In L. White and B. Tursky
(Eds.), Clinical biofeedback Efficacy and
Mechanisms. Guilford. U.K.
Orne. M.T. (1982) Perspectives in biofeedback: ten years
ago. today, and In L. White and B. Tursky (Eds.),
Clinical biofeedback Efficacy and Mechanisms.
Guilford, U.K.
Riley, D.M. and Furedy, J.J. (1981) Effects of
instructions and contingency of reinforcement on
the operant conditioning of human phasic heart rate
change. Psychophysiology. 1X: 75-81.
Riley. D.M. and Furedy, J.J. (1982) Reply to
Mulholland. In L. White and B. Tursky (Eds.).
Clinical biofeedback Efficacy and Mechanisms.
Guilford, U.K.
Schober, R. and Lacroix, J.M. (1986) Effects of task
instructions and contingency on the development of
phasic heart rate control and its correlates. Curl. J.
Psychol. 40: 54-64.
Stern, J.A. (1964) Towards a definition of
psychophysiology. Psychophysiology, 1: 90-91.
Tursky, B. (1982) An engineering approach to
biofeedback. In L. White and B. Tursky (Eds.),
Clinical biofeedback Efficacy and Mechanisms.
Guilford. U.K.