The present study was designed to evaluate the efficacy of utilizing an avoidance conditioning paradigm in EMG biofeedback training and to compare this method to the standard biofeedback training paradigm. Frontalis EMG levels of 20 college students were monitored during non-stress and stress conditions. Half then received standard EMG biofeedback training. The other half received biofeedback with contingent aversive stimulation. Both groups received training to a relaxation criterion of 3 microvolts for 100 seconds or, for a maximum of two 20 minute sessions. Subjects were then monitored again during non-stress and stress conditions. Both groups obtained significant EMG reductions due to training with no significant differences between them. Standard biofeedback training required less time for subjects to achieve the relaxation criterion than did biofeedback with a shock-avoidance contingency. Possible applications of avoidance contingent biofeedback were suggested.

This study was designed to compare the clinical effectiveness of EMG biofeedback and progressive relaxation training in the treatment of muscle tension headache. These procedures also were compared with a treatment-element control group. Results from this study indicated that EMG biofeedback, progressive relaxation, and the control procedures all led to significant improvements across sessions on EMG and most self-report measures. There was little evidence that either treatment technique was superior to the other or to the control procedures. Although in most cases there were rather large numerical differences between groups, these differences generally were not statistically significant. Analysis of correlations between EMG and self-report data revealed a pattern of variable but generally nonsignificant relationships. However, for the biofeedback and progressive relaxation groups, there were a number of highly significant correlations. The pattern of correlations suggested that the relationship between EMG tension and subjective headache pain may be better predicted by something other than a strict linear model.