Electroencephalographic analysis: A methodology for evaluating psychotherapeutic process

323

Psychiafry Research, 2, 323-329 (1980)

0 Elsevier/North-Holland

Biomedical

Press

Electroencephalographic Analysis: A Methodology Evaluating Psychotherapeutic Process Murray Alpert, Received

Neal L. Cohen,

Merrill

Martz,

for

and Carlan Robinson

February 26, 1980; accepted April 21, 1980.

Abstract. The relationship between individual differences in listening style and clinical sensitivity was studied. Listening style is conceptualized as extending from a critical, analytic, focused attitude to a holistic, intuitive, free-floating attitude. We studied the listening styles of psychiatric residents while they listened to a tape recording, a 17-minute fragment of a psychotherapy that had been rated by a panel of experts as containing 22 cues reflecting the patient’s concern with termination of treatment. The ability of clinicians to identify these cues is our experimental measure of clinical sensitivity. Our measures of listening style include eyeblink rate, memory for high-imagery words, and electroencephalogram recorded from electrodes over the right and left temporal areas, all measured while psychiatric residents listened to the tape. Results suggested that an abundance of alpha indicates a readiness on the part of the listener to process clinically relevant information. In addition, a listening style with engagement of the right hemisphere predicted clinical sensitivity. These findings suggest ways to study how the skilled clinician “tunes” himself empathically to his patient. Key Words. Psychotherapy, rhythm.

clinical sensitivity,

lateralization,

hemisphere,

alpha

In this report we explore an analogue situation for the study of the way psychotherapists process clinically relevant information. Since Freud’s early papers, clinicians have favored a “free-floating” listening style as a facilitator of empathy. The components of this listening style are elusive, and we attempt to capture it through study of its cognitive and physiological correlates. In previous studies we found significant relationships between listening style and the clinician’s sensitivity to the termination issues in an analogue therapeutic situation (Alpert and Cohen, 1979; Cohen and Alpert, in press). We examined two correlates of listening style, one physiological (eyeblinks), and one cognitive (memory for high-imagery words). “Listening style” referred to the directedness of the listener’s attention; at one pole was a focused, analytic attitude, an active attempt to extract information. At the other extreme was a free-floating attitude, a “listening away” from the focus of what was being said. Research in neuropsychology (Sperry, 1973) indicates that the two cerebral hemispheres are differentially specialized. The dominant hemisphere, usually the left, excels in verbal, analytic, logical, deductive, evaluative tasks, but does relatively less well with visual, spatial, holistic, emotional modes; the opposite is true for the

Murray Alpert, Ph.D., is Professor; Neal L. Cohen, M.D., is Assistant Professor of Clinical Psychiatry; Merrill Martz. Ph.D., is Clinical Instructor; and Carlan Robinson, Ph.D., is Postdoctoral Fellow, Department of Psychiatry, New York University School of Medicine, 560 First Avenue, New York, NY 10()16. (Reprint requests to Dr. Alpert.)

324 nondominant hemisphere. Further, the differential activity of the two hemispheres may be reflected in the electroencephalogram (EEG) recorded from their temporal areas, with alpha waves (8-13 Hz) indicating a resting, idling state, while desynchronization of alpha is taken as indicative of activation of the underlying hemisphere (Doyle et al., 1974). Mackworth (1968) has also provided a model consistent with the suggestion that the presence of abundant alpha activity reflects the engagement and unavailability of the individual. Thus attentional focus, as measured by alpha level following receipt of the significant information, may be potentially relevant to sensitive clinical detection. Spence and his co-workers were interested in devising an experimental paradigm to examine Freud’s (1958, p. 111) suggestion to listen to patients with “an evenly divided” attention. They reported several studies using calibrated clinical samples to assess the success of undergraduate subjects in clinical detection tasks (Spence and Grief, 1970; Spence and Lugo, 1972). These studies suggest that it is the way the attention is distributed rather than the total amount of attention that determines whether certain kinds of information can be processed. Interest in “clinical listening” has led to investigations of where and under what circumstances the right and left hemispheres of the brain differentially process information. The learning studies of Paivio et al. (1968) and Paivio and Yuille (1969) have, in fact, suggested information processing differences in listening style. These authors have proposed differences in the encoding process for high- and low-imagery words. High-imagery words are seen as encoded with both a verbal identifier and a visual image identifier, which would involve both the left and the right hemispheres, while low-imagery words are encoded with only a verbal identifier. Further, subjects with relatively better memories for high-imagery words tend to mediate cognitive tasks with imagery, suggesting utilization of both the right and left hemispheres. Gale et al. (1972), in a study of imagery and individual differences in the capacity for vivid imagery, found that imagery, per se, leads to a reduction of alpha level. Highimagery words elicited more vivid imagery than low-imagery words for both weak and vivid imagers. The differential effect of stimulus imagery was greatest for the vivid imagers. The EEG waves of weak and vivid imagers, under eyes open, minimal demand conditions, differed in mean dominant frequency of alpha rather than in alpha abundance. Gale et al. (1972) reported greater alpha suppression by highimagery words than by low-imagery words. Voluntary imagery resulted in opposite effects, indicating that increasing the effort involved leads to greater suppression of alpha abundance. Thus, the type of imagery, as well as the facility of the individual, is reflected in the differential effect on alpha level; the vivid imager consistently appears to be more involved, as indicated by ratings of imagery and reflected in alpha level. Putting these findings together, Mackworth (1968) has provided a general neurophysiological model of the response, in terms of alpha level, of the brain to external information. The work of Spence and his co-workers (Spence and Grief, 1970; Spence and Lugo, 1972) has suggested that rather than the amount of attention specifically, it is the way in which attention (or neurological responding, in physiological terms) is distributed that is relevant to clinical listening. Paivio et al. (1968) and Paivio and Yuille (1969) have further refined our understanding of hemispheric variation that is involved in selectively responding to varying kinds of information. Gale et al. (1972)

325 have replicated and extended the learning studies of Paivio et al. by providing further empirical evidence of the differential interaction of information and individual, in terms of alpha levels, observable behavior, and individual self-reports. Our interest in this study was to extend those findings to the analogue clinical situation in an attempt to further our understanding of clinical detection. In the study to be reported here, EEG was recorded from electrodes over the right and left temporal areas while psychiatric residents listened to a 17-minute fragment of a psychotherapy session. Doyle et al. (I 974) had found that the left/ right ratio of alpha band activity (8- I3 Hz) of EEG was sensitive to the engagement of a subject who was involved in tasks requiring primarily verbal or spatial cognitive mediators. Alpha tended to be greater from the hemisphere presumed not to be involved in a particular task. Thus, it is anticipated that study of brain asymmetry and changes in the alpha state in this study would elucidate the ways in which the clinician “tunes” himself to the patient’s communication, thus making himself ready to receive and to process complex information about his patient. Methods Subjects. Seventeen second and third year residents in psychiatry at the New York University/ Bellevue Medical Center served as subjects. Each subject sat in a lounge chair in a quiet treatment room and listened to instructions, segments of poetry, and the psychotherapy fragment (described below). The Tape. The 17-minute tape-recorded simulation of a woman speaking in a psychoanalytic session was played individually to each subject through a loudspeaker while the subject sat in a lounge chair in a quiet treatment room. The tape is a compilation of clinical segments in which a woman discusses childbirth and weaning, and then recalls a dream in which she is left on a platform while a train departs from the station. Calibration of the recording for references to the theme of termination had been done independently by three experienced psychoanalysts who listened to the recording while they read through transcripts. We confirmed Spence’s calibration, with the addition of one cue found by a majority of our listeners. Twenty-two statements were agreed to be directly or indirectly related to the theme of termination of treatment. Many of the cues are direct but embedded in a context that may conceal this aspect of their meaning (e.g., the patient mentions Independence Day, and discusses a trip to the town of Liberty in Kansas). Other termination cues are more symbolic, using metaphor or analogy (e.g., How long do you nurse the baby?). Procedure. Each subject was tested individually. EEG was recorded from temporal areas referred to a common vertex lead. Band-pass filtering eliminated nonalpha activity, and as had been done in our earlier studies (Alpert and Cohen, 1979; Cohen and Alpert, in press), extraocular electrodes were attached to measure eye movements and blinks. EEG was measured from temporal areas (Pj and P4) referred to a common Cz lead. These signals and an electrooculogram and audio channel were recorded on FM tape. Alpha activity was extracted from EEG with a matched pair of 8-13 Hz band-pass filters whose output was sampled by a laboratory computer that computed

326

the variance (power) for every 500-msec epoch. Subjects first listened while a brief poem was played twice. Before the second playing of the poem, subjects were asked to estimate the poet’s age, sex, and nationality. Another poem was played, and instructions were varied to attempt to shape subjects into one or another of two listening styles. On a randomized schedule, half the group was instructed in listening with a free-floating, noncritical attitude while the other half was instructed in focused, analytic attention. This listening instruction was next applied to the therapy segment. In addition, each subject was primed to the termination theme and asked to attend to references to that theme. Our purpose was to study individual differences in listening style and their relationship to measures of clinical sensitivity. While EEG was our psychophysiological measure of listening style, we sought a cognitive measure of listening style by testing our subjects for their memory of high- and low-imagery words embedded in the therapy tape. All listeners were tested for recognition of 33 nouns which had been spoken by the patient. Each of these words was matched by two filler nouns not spoken by the patient, equal in number of syllables, initial letter, and frequency in the English language as recorded in the Thorndike-Lorge Wordbook (1944). The filler nouns were selected for imagery evocation value as reported by Paivio et al. (1968). Rated on a 7-point scale, the 33 high-imagery filler words had a mean imagery value of 6.52, while the low-imagery word fillers had a mean imagery value of 2.75. All 99 nouns were then randomly sorted and recorded on tape played through the loudspeaker. After hearing each word, the subject was asked to respond “yes” or “no” according to whether the word had been spoken. A d’score (Kintsch, 1970), a measure of the rate of recognition of stimulus words corrected for the rate of false-positive responses to filler nouns, was computed separately for high- and low-imagery words. Miller (1972) and Paivio et al. (1968) suggested that subjects with relatively higher d’ for high-imagery words tend to mediate cognitive tasks with imagery. This would correspond to our conception of the “free-floating” listening style frequently prescribed for clinical listening. Results Eight subjects were instructed in a free-floating listening style and nine as focused analytic listeners. The groups did not differ in either blink rate, d’ for high-imagery words, or ratio of alpha EEG energy in the left compared to the right temporal leads. We were not able to reliably redirect the subjects’ polar listening styles. Half the group had been instructed to listen with a free-floating, noncritical attitude, while the other half was instructed in focused, analytic attention. The two correlates of listening style that we had used in the past (Alpert and Cohen, 1979) indicated that the instructions did not shift the individual’s intrinsic listening style. Similarly, the EEG revealed a baseline configuration of brain activity which varied little under our experimental procedures. Our eyeblink measure was not a significant correlate of sensitivity to the “termination” theme. Regardless of the listening plan assigned, however, our cognitive measure of listening remained a significant correlate of successful cue recall (r = 0.72, p < 0.01).High-imagery language recognition favored recall of statements relevant to the “termination’* theme. Similarly, the EEG asymmetry measure (the percent of left-sided to left-plus-right alpha energy) correlated with the number of cues detected

327 (r= 0.68, p < 0.01). These results extend our previous finding that a listening style suggesting involvement of the right hemisphere is associated with clinical sensitivity. Within-subject analysis of the EEG measures was also done. For this analysis, EEG activity of subjects listening to cues that were later detected was compared with activity during cues that were missed. All seven subjects who had a sufficient number of recalled cues and complete EEG during the cues were used in this analysis. For the group of seven subjects, a main effect was an increased density of alpha activity over the 2- to 3-second period just before the onset of cues that were later recalled by subjects (F = 34.1; df = 1,210; p < 0.001) as compared to that just before the onset of cues that were not recalled. Differences in alpha density persisted during the cue phase, although not so sharply as in the period just before the onset of cues (see Fig. 1). this would suggest that abundance of alpha indicates a readiness on the part of the listener to process clinically relevant information. Looking next at alpha asymmetry data, we found that at least some of the residents showed EEG hemisphericity shifts that coordinated with successful cue detection. In a second-by-second analysis of left/right alpha ratio, detected cues were compared to missed cues among the group of seven residents. There was a trend toward a divergence between hits and misses, so that at 2 seconds following the onset of the cue the percent of left alpha for the recalled cues was significantly higher than that for the missed cues (F = 4.50; df= 1,24; p < 0.05). This would suggest that with input of the clinical information, optimal modes of hemispheric cross-talk facilitate the processing and recall of the information. Since our cues differed in duration, time-locked analysis must be interpreted cautiously.

Fig. 1. Alpha abundance unrecognized

at onset of termination

cues recognized

and

-.4 -.5 I -2.5

I -2

I -1.5

I -I

I -0.5

0

I 0.5

I I

I 1.5 TIME

Period

prior

to cue

@Cue

begins

I 2 IN

I 2.5 SECONDS

328 Comment These results suggest that for optimal sensitivity to the patient’s concerns with termination, a listening style with increased alpha density brought about a state of heightened receptivity to the relevant cues. The cues differed from each other in their directness, in their surplus meaning, and in their imagery. It appears that the more metaphoric cues are understood with the processing style characteristic of freefloating attention deployment. Not all cues seemed to require a free-floating listening style, and the data suggest that study of differences in clinical detection may have a relationship to processes involved in the patient’s production of cues. Are there patterns of EEG hemisphericity in the patient which are conducive to cue production? Is cue production indicative of therapeutic effort? In proposing to study these issues, we hypothesize that the study of the interaction between modes of attention deployment in patient and therapist, their rapport, and their significant communications may be accessible to these methods. If our findings can be replicated and validated, it may lead to a better understanding of the way that skilled clinicians process large amounts of information. Supporting our belief is a recent study by Hasenfus et al. (1978). They found that graduate psychology students who responded like type A therapists on the A-B scale showed significantly more alpha while listening to clinical segments and while relaxing than did their B student counterparts. Some studies have found A’s tend to be more effective therapists than B’s with schizophrenic patients (Betz, 1963; Berzins et al., 1972), while other studies suggest that A’s are effective with all patient types (Chartier and Weiss, 1974) supporting our finding that the alpha state identified the more sensitive clinician listeners. A primary goal of this study was the development of systematic approaches to clinical training. At present, because they are subtle and difficult to specify, clinical of our skills are acquired laboriously in a “hit or miss” fashion. The application electrophysiological techniques to calibrated clinical samples may make clinical training more systematic, and allow us to begin to learn how the skilled clinician “tunes” himself empathically to his patient. References Alpert, M., and Cohen, N. A clinical listening analogy for study of listening style in psychiatric residents. Journal of Psychiatric Education, 3, 36 ( 1979). Berzins, J.I., Ross, W.F., and Friedman, W.H. A-B therapist distinction, patient diagnosis and outcome of brief psychotherapy in a college clinic. Journal of Consulting and Clinical Psyhologv.

38, 23

1 ( 1972).

Betz, B.J. Differential success rates of psychotherapists with “process” and “non-process” schizophrenic patients. American Journal of P.yvchiatrv, 119, 1090 (I 963). Chartier, G.M., and Weiss, L. A-B therapists and clinical perception: Support for a “Super A” hypothesis. Journal of Consulting and Clinical Psychology,. 42, 3 I2 (1974). Cohen, N., and Alpert, M. Styles of listening and clinical sensitivity. Archives elf’ General Ps~~chiatr~~(in press). Doyle, J., Ornstein, R., and Galin, D. Lateral specialization of cognitive mode: II. EEG frequency analysis. Ps_vchophysiologv, 11, 567 ( 1974). Freud, S. Recommendations to physicians practicing psychoanalysis. (19 19) Standard Edition. Vol. 12. Hogarth Press, London (I 958).

329

Gale, A., Morris, P.E., Lucas, B., and Richardson, A. Types of imagery and imagery types: An EEG study. British Journal of Ps.r’chology, 63, 523 (1972). Hasenfus, N., Martindale, C., and Kaplan, C. Alpha amplitude in A & B type clinical and experimental psychologists. Psychotherap.v: Theory, Research and Practice, 15, 126 (1978). Kintsch, W. Learning, Memory and Conceptual Processes. John Wiley & Sons, Inc., New York (1970). Mackworth, J.F. Vigilance, arousal, and habituation. fsychological Review,, 75, 308 (1968). Miller, T. Some characteristics of two different ways of listening. Unpublished doctoral dissertation, New York University Graduate School (1972). Paivio, A., and Yuille, J.C. Changes in associative strategies and paired-associate learning over trials as a function of word imagery and type of learning set. Journal of Experimenfal fsvchology,

79, 458

(I 969).

Paivio, A., Yuille, J., and Madigan, S. Concreteness, imagery, meaningfulness values for 925 nouns. Journal qf Experimental Pswhology, 76 (Part 2) I ( 1968). Spence, D.P., and Grief, B. An experimental study of listening between the lines. Journal of Nervous and Mental Disease, 151, 179 (1970). Spence, D.P., and Lugo, M. The role of verbal cues in clinical listening. In: Holt, R., and Peterfreund, E., eds. I?v,vchoana/y.sis and Contemporar,~ Science. MacMillan Company, New York (1972). Sperry, R.W. Lateral specialization of cerebral function in the surgically separated hemispheres. In: McGuigan, F.J., and Schoonover, R.A., eds. The P.~ychophysiolog~~of Thinking. Academic Press, Inc., New York (1973). Thorndike, E., and Lorge. 1. The Teachers Wordhook qf’30,OOO Words. Teachers College, Bureau of Publications, New York (1944).