After the polygraph, biofeedback is arguably psychophysiology's 2nd most salient purported application. My position on evaluating any treatment based on psychological or psychophysiological principles is that for an evaluation to be systematic and scientific, rather than anecdotal or pseudo-scientific, the logic of evaluation must adhere to a specific effects (SE) approach.
The logic of the SE approach is accepted in such applied areas as pharmacology. To evaluate the claim that aspirin lessens headaches, one must run a placebo-control comparison where two conditions are compared which differ only with respect to the administration of the drug aspirin. The comparison must be "double blind" to rule out the possibility that the condition in which the aspirin is administered shows an improvement only because the subject believes that the aspirin will be helpful. The double-blind control involves the "control" condition being one where neither the subject nor the experimenter knows that the subject is getting a drugless pill. It should be noted that establishing that aspirin has a beneficial effect on head aches is not the same as establishing what the mechanisms are that mediate that beneficial effect. In the case of aspirin, there is a host of established beneficial effects where the mechanisms of such effects is still unknown.
Still, research that establishes the conditions under which aspirin has beneficial effects is likely to lead to an increase of knowledge concerning the mechanisms involved, as well as improvements in those beneficial effects. In addition, by observing those different conditions (which include individual differences), science can also aid technology by identifying those conditions where aspirin has no beneficial effects, or even where it has detrimental effects. It is for this reason that science-based technologies show progressive improvements in efficacy, whereas pseudo-science-based technologies like the administration of snake oil show no such progressive improvements, being dependent solely on faith in the power of the administrator of the treatment.
Most psychological treatments are not classifiable in simple objective terms that indicate the essential way or ways in which they differ form other psychological treatments. For, for example, it would be difficult to evaluate the specific beneficial effects of cognitive behavior therapy is only because the cognitive and behavioral approaches have elements that are contrary to one another, and also because, even though abbreviations such as CB are employed, it is quite clear that a CB therapy in the hands of one operator may be quite different in the hands of another. This is in stark contrast with pharmacological treatments, where aspirin does not differ as a function of who prescribes it or who administers it.
However, biofeedback treatment is as specifiable as objectively as aspirin is. The essential claim of biofeedback therapy is that information about moment-to-moment changes in physiological functions of which we are not conscious will have a beneficial effect on enabling us to increase control over those functions. So, for example, in the treatment of essential hypertension patients were provided with information that may have involved telling them to avoid smoking or violent arguments with their spouses, because these behaviors were likely to increase blood pressure, and blood pressure may have been monitored once a month, with the target behavior (TB ) being defined as the lowering of blood pressure. In information-processing terminology, the pre-biofeedback control of blood pressure was a feedforward treatment, as exemplified by a GP telling patients that if they got into heated arguments, their blood pressure would increase, with the hope that providing this feedfoward information (i.e., instructing them) would result in an improved TB of lowering blood pressure. It will be noted that this feedforward, instructional treatment required no psychophysiological equipment (hardware) or expertise (software) from the therapist.
In contrast, a biofeedback treatment of blood pressure would involve providing the patient with much more "dense" information about his or her blood pressure, with this information being as "dense" as being provided every 5 seconds (possible if one uses a finger transducer, which may not be accurate for actual blood pressure, but reflects changes accurately). It is the provision of this "dense" information that, according to the biofeedback assumption, is beneficial in increasing TB (here the lowering of blood pressure).
Clearly, in terms of the SE approach that is common for evaluating pharmacological treatments, the appropriate control condition is one where the information is equally "dense", but where it is not as accurate a reflection of actual blood pressure changes as in the experimental condition. The logic of pharmacological evaluation used to be well known to experimental psychologists in the fifties and sixties. The issue that concerned those researchers was whether a non-biologically significant even like the onset of a weak light could act as a reinforcer for operant, "Skinnerian" conditioning, where the target behavior (TB) was an increase in the rate of pressing a bar by rats, and the reinforcement was not food or water or even sex, but merely the onset of the light. The first studies that were used to demonstrate that light was reinforcing were ones where the experimental group received a light contingent on a bar press, and the control group did not. The results showed a higher rate of bar pressing (more of the TB) in the experimental group, but this was not accepted by the research community as adequate evidence. It was necessary to show, as was done, that it was the contingency between bar pressing (the TB) and reinforcement (or feedback) that was responsible for the increase in bar pressing, rather than some non-specific, motivating effect of the light. The method used was the so-called "yoked-control" method, wherein the experimental or "master" rats produced the lights through their bar pressing, while the control or "yoked" rats received lights on the basis of the bar pressing of the master animals. In other words, the control condition removed the contingency between the TB and the reinforcer (or feedback), while in all other respects the two conditions were the same.
Even though Neal Miller, arguably the most eminent experimental psychologist at the time, was familiar with the SE-based manipulation-of-contingency control, in his famous 1969 science article (http://www.sciencemag.org/content/vol163/issue 3866/index.dtl), he employed the so-called "bidirectional" control in which, for example, he assessed the effect of biofeedback on heart rate deceleration as the TB by comparing it with a condition where the TB was heart rate acceleration.
In methodological terms this is like establishing that light reinforces bar pressing in rats by using a contingency where non-bar pressing results in the provision of light as the control condition. Because non bar pressing (without previous training) is a more probable behavior than bar pressing, one could easily get apparently "positive" results not because bar pressing was being reinforced, but because the competing response of non-bar pressing was being reinforced. Similarly, heart-rate acceleration is easier to produce than heart-rate deceleration (i.e., it is easier to get excited than to calm down), so that the "bidirectional control" comparison is not a valid method for establishing a specific beneficial effect of biofeedback over the control of heart-rate deceleration (which is the medically desirable TB). So if the target behavior is (the medically desirable, but hard to achieve via feedworard) heart-rate deceleration (note that in the later human heart-rate lab studies, the typical heart-rate decelerations achieved through biofeedback continued to be around 1 bpm, which is the same as that achieved by using feed-forward repetition of a mantra), then no bidirectional control established that biofeedback has a specific, enhancing effect on increasing the production of the declarative target behavior. It should be noted that feedworward-based target behavior is subject to normal learning principles, but this improvement in the target behavior is due to feed-forward, instructional leaning (that occurs in any psychotherapy or any other sort of advice given by one person to another), and not biofeedback learning.
The magnitude of the effects reported by Miller in "dumb" curarized rats, together with one spectacular demonstration of within-subject bidirectional control (where rats could apparently dilate and constrict in each ear depending on whether the ear received feedback for dilation or constriction) tender to render his failure to use the SE contingency-control method empirically unimportant. However, later work, including by Miller himself, showed that his 1969 results were not replicable.
Nevertheless, in the meantime, many research labs and clinicians took up biofeedback with enthusiasm, and most not only did not use the contingency-control approach, but even employed the more conceptually primitive no-feed back control. This parallels the early light-reinforcement studies for bar pressing which employed a no-light control, or a pharmacological test of the efficacy of aspirin that employed a control condition where subjects were not given any pill at all. Perhaps one reason why even the research community seldom employed the proper SE controls for biofeedback (see, e.g., Furedy, J.J. (1987). On some research-community contributions to the myth and symbol of biofeedback. International Journal of Psychophysiology, 4, 293-7.) is that Miller had obtained such spectacular results with his curarized rats, that it seemed obvious that intact college students could do the same. When the replication problems (reported by Miller himself) emerged, few in the research community, and fewer still in the clinical community took any notice.
One result of this is that the possibility of science-based improvements for those specific conditions where biofeedback did, in fact, have a specific beneficial effect if administered properly, disappeared. For example, the EEG alpha feedback area died a quick death at least in the research community, but ignored the "chunking" method which showed a specific beneficial effect in producing alpha waves against a proper non-contingent control (http://www.ci.nii.ac.jp/ naid/110003162432/en/).
Here you can
Website Designed by