An EEG biofeedback protocol for affective disorders.

Abstract

INTRODUCTION Despite our own older work which showed that biofeedback with somatosensory evoked EEG potentials can have profound effects on pain perception in rats and it seemed to us quite a leap to think that an application of EEG biofeedback could be effective in treatment of depression and other affective disorders. Yet the logic of developing such an intervention would be no different than that which was used in our work in the pain modality: In that work, we were aware of-a sizable literature documenting evoked EEG potential (EP) correlates of We reasoned simply that if a large value of a particular EP component accompanied intense pain, whereas a smaller value accompanied no pain or analgesia, then if one could train individuals to reduce the particular EP, one ought to see reductions in perceived pain. This approach yielded very promising Thus it seemed to us that a reliable EEG index of affect could be found, then we would be in a position to develop an EEG biofeedback protocol for depression. However, until the relatively recent publication of work from R. J. Davidson's laboratory,g there were no documented reliable indices of affect in the waking EEG. Based on evidence from the neurology literature, Davidson and associates hypothesized that the right frontal cortex contained a neural system mediating negative emotion and avoidance behavior, whereas, in contrast, the left frontal cortex contained a neural system mediating positive affect and approach behavior. An active cortex is known to show higher (13-30 Hz) "beta" frequencies in a low amplitude, desynchronized EEG, whereas an idling or inactive cortex is known to show lower (8-12 Hz) "alpha" frequencies of synchronous (sinusoidal) higher amplitude activity. Davidson and colleagues thus hypothesized that positive emotion should correlate with high beta and low alpha activity in the left frontal cortex and with low beta and high alpha activity in the right frontal cortex. Negative emotion would correlate with the reverse pattern of cortical activity: high left frontal alpha, low lefi frontal beta, high right frontal beta, and low right frontal alpha. Because there are harmonics of electromyographic activity reaching down to the beta range, which could be therefore mistaken as beta, many researchers have focused on the alpha (inverse) indices of emotion. (It is possible to utilize beta, but it requires added steps to correct for electromyographic artifact.) In a series of ingenious, original experiments, Davidson and colleagues provided a strong set of evidence that cortical activation asymmetry as inversely indexed by alpha power or magnitude, was a reliable correlate of positive and negative emotion. The asymmetry metric developed by the Davidson group will be referred to here as the asymmetry score =A, = log R log L where R is alpha power at cortical site F, and L is alpha power at cortical site F, . It is also possible to define an asymmetry score as A, = (R-URtL). Although A, and A, are not mathematically equivalent, they correlate very highly (2 .98).lD1' What has been generally found is that higher A, or A, scores go with positive affect and lower A, or A, scores go with negative affect. (Hereafter, the unsubscripted term "A-score" will be used to refer to generic alpha asymmetry indexed by either A, orA2). The former condition means relatively greater left frontal activation; the latter means relatively greater right frontal activation. The term "relatively" is used here because in any individual case, one cannot say from an Ascore whether the critical effects are in the left versus right cortex, or both, since A-scores combine R and L. Davidson and co11eagues9 did a variety of studies to support their hypothesis. For example, they showed that a person's resting frontal alpha asymmetry predicted their affective responses to emotionally positive and negative film clips.12 They also showed that rewards and punishments led to differential asymmetry response^,'^ and showed that facial expressions of emotion were systematically related to asymmetry scores.914 Most relevant to this article, (1) Henriques and Dav ids~n '~ in 1990 showed that currently depressed persons have left frontal hypoactivation (lower A,-scores) in comparison with never depressed persons: (2) Henriques and Davidson16 in 1991 showed that previously depressed but now remitted persons show also a relative left frontal hypoactivation in comparison with never depressed persons. Gotlib, Ranganath, and Rosenfeld" in 1998 replicated and extended this finding, in one study, by comparing three groups: currently depressed, formerly depressed,