Fredric Schiffer, M.D.
Martin H. Teicher, M.D., Ph.D.
Andrew C. Papanicolaou, Ph.D.

From the Department of Psychiatry, Harvard Medical School, and the Developmental Biopsychiatry Research Program, McLean Hospital, Belmont.  Drs. Schiffer and Teicher are at Harvard and McLean; Dr. Papanicolaou is at the Department of Neurosurgery, The University of Texas Medical School, Houston

Address for Reprint Requests:  Fredric Schiffer, M.D., McLean Hospital, 115 Mill Street, Belmont, MA 02178

Running Title:  Evoked Potentials During Memory

Key Words:  Probe Auditory Evoked Potentials, Psychological Trauma, Cerebral Dominance

Corresponding Author:  Fredric Schiffer, M.D., McLean Hospital, 115 Mill Street, Belmont, MA 02178, (617) 237-9620.

The authors thank Scott Lucas, Ph.D., and Kenneth Levin, M.D., Ph.D. for their comments and suggestions, and Yutaka N. Ito, M.D., Ph.D. for blind rating of patients.  The research was supported, in part, by gifts from the Hall Mercer Foundation, Snider Family, and NIMH grant MH-48343 (MHT).

Submitted to: The Journal of Neuropsychiatry and Clinical Neuroscience
     May 16, 1994. Revised August 22, 1994. Revised December 5, 1994.     Accepted January 5, 1995.


 Auditory probe evoked potential attenuation was measured as an index of hemispheric activity in 10 subjects with a history of childhood trauma and in 10 matched subjects without, while they recalled a neutral memory and then a traumatic memory.  There were prominent group differences in degree of cerebral laterality between memory tasks (p = 0.02).  The victim group had a significant left dominant asymmetry during the neutral memory (p < 0.02), which markedly shifted to the right during the unpleasant memory (p = 0.007 for degree of shift).  Normal controls did not display a significant asymmetry during either task, nor a significant shift between tasks.

 Although there has been a long history of speculation about the mental properties of the right brain (1), it was not until Sperry's (2) landmark work on post commissurotomy patients that objective studies on the functions of the right brain accelerated.  Among its properties, the right hemisphere has been shown to be involved in the processing of affect.  For instance, the right side has been shown to be superior at the perception and recognition of emotion (3-7).  Weschler (8) and Masters (9) have shown that patients with right sided lesions have impaired recall of affective memories.  Several studies have indicated right sided involvement in the experience of emotion, particularly unpleasant emotion (10-16).
 The present study represents our initial effort in subjects with a history of trauma to obtain evidence about the functioning of right hemisphere in a setting resembling a psychotherapy session in which unpleasant early memories are affectively recalled.  This condition was compared with one in which the subject recalled a recent neutral, work-related situation.  The responses of these subjects with a history of trauma were compared to those of a control group without a history of significant trauma.
 As a measure of hemispheric activity, we used probe auditory evoked potentials (AEPs) as developed by Papanicolaou (17).  The method entails exposing each subject to repeated auditory clicks and measuring, simultaneously over both hemispheres, the amplitude of the averaged evoked potentials in response to the clicks.  While exposed to the clicks, the subject is asked to engage in a mental activity.  When one hemisphere is more involved than the other in a cognitive activity, then the more involved hemisphere is expected to display a relatively attenuated evoked response to the clicks.  Papanicolaou and Johnstone (17) have reviewed the theory, methods, and applications of the probe AEP.

 Twelve paid volunteers were recruited who believed they had come from a dysfunctional family, were right handed, over the age of eighteen, and on no medication with known psychotropic effects.  Twelve similar subjects who believed they did not come from a dysfunctional family were also studied.  After a full explanation of the procedures, written, informed consent was obtained.  A psychiatrist then conducted a full psychiatric interview and mental status examination.  No subject had to be excluded because of an active DSM-IIIR Axis I or II disorder.  Right handedness was confirmed by the Edinburgh Handedness Inventory (18).
 Electrodes were applied at C3 and C4, using a lycra cap (Electro Cap) and referenced to linked ear electrodes, and to a forehead ground contained in the cap.  Ten mm gold electrodes were applied lateral to and below the right eye to monitor conjugate eye movements and blinks.  All impedances were below 5 K ohms and equal bilaterally to within ±1 K ohm.  The subjects were asked to sit back in a reclining chair with a rolled towel used as a neck support.  The patients fixed and maintained gaze on a mark in front of them throughout each recording period and were closely watched for eye movements.
 The subjects were first asked to remember and reflect on a recent ordinary work or school situation.  They were asked to raise their right hand at the wrist at the start, and to lower it when they were actively remembering the situation.  If they were no longer engaged in the activity, they could signal this by raising their hand again.  The recording of AEPs commenced when they lowered their hands.
 Evoked potentials were recorded on a QSI-9000 computerized EEG set to produce binaural 86 dB clicks (3 per second) and to record evoked responses for 250 msec after each click.  Three to six hundred epochs were averaged. Between groups there was no statistical significant difference between the number of epochs recorded or rejected.  If it appeared that a readable evoked potential response had not been obtained, the recording was repeated.  The low-frequency EEG filter was set at 1 Hz, the high-frequency filter at 30 Hz.  Epochs greater than 15.5 µV  were automatically rejected as possible artifact.
 Following the recording, the subject was given several queries taken from the POMS scale (19) to monitor his affect.  Specifically, he was asked to measure, on a 5 point scale from none to extreme, his level of tension, anger, sadness, hopelessness, nervousness, panic, and guilt.  Subsets from the POMS scale have been used as measures of subjective mood (20).
 A psychiatrist then engaged the subject in an empathic psychiatric interview lasting about 15 minutes in which he was asked about his early family life.  The psychiatrist tried to affectively engage the subject, and to get him to share, with emotion, a painful childhood memory.  When the psychiatrist felt that the subject was affectively reexperiencing the memory, he asked him to try to continue to maintain his memory and mood, but without speech or motion, so that his evoked potentials could be measured.  Following the recording, the abbreviated POMS scale was again used to measure his emotional state.  The unpleasant memory task was always presented after the neutral memory task because of concern that the lingering effects of the unpleasant memories would interfere with the neutral task.  Following completion of the study, the psychiatrist worked with each patient to restore his usual mood, and no subject left the laboratory in distress.
 The averaged AEP response from each condition was printed, and all recordings were blindly read by an experienced research electroencephalographer to obtain N1 and P2 peaks.  N1 was defined as the maximal negative deflection between 70 and 130 msec which conformed to expected patterns, while P2 was defined as the peak of the following positive wave.

 For each task an asymmetry index, (C3-C4)/(C3+C4), was calculated from measurements of the N-P amplitudes at C3 (left auditory cortex) and C4 (right auditory cortex).  Probe AEP's are not localizeable within a hemisphere and are most reliable when recorded over the primary auditory cortex since that location reduces the contaminating effects of volume conduction.  Group differences were evaluated using ANCOVA with right/left hemisphere response as a within subject factor.  Covariate procedures were used to provide partial statistical control for group differences in degree of emotional response to the memory conditions.  T-tests were used to test the null-hypothesis that there was no hemispheric asymmetry in the evoked potential response.  Data are presented as mean ± S.D.  Probability values were determined using two-tailed tests.

 Of the 24 subjects studied, 20 had recordings over each hemisphere and during each task which had clearly discernible N1 and P2 peaks.  Four subjects had to be eliminated because at least one of their 4 tracings did not show a clearly defined N1 or P2 peak.  It is not usual in evoked potential research for a percentage of recordings to be uninterpretable.  Our data were collected and then later read blindly to subject's identity.  Our expert reader (AP) found 91 of the 96 evoked potentials recorded to be clearly interpretable.  The five recordings which had ambiguity belonged to the four subjects (2 from each group) who therefore were eliminated.
 Ten subjects (mean age = 32.9 yrs; M5/F5) reported at the time of recruitment that they had experienced significant childhood trauma.  Another group of 10 subjects (mean age = 33.0 yrs; M4/F6) reported at the time of recruitment that they did not have significant childhood trauma.  All subjects had completed at least 2 years of college.  At the time of the study a psychiatrist interviewed each subject and wrote up a history for each individual.  Two other physicians blindly evaluated each history and subjectively rated the degree of each subject's early psychological trauma on a 5 point scale from 0 (= no known abuse) to 4 (= extreme abuse) for six categories of trauma.  For overall trauma there was a high correlation between the 2 raters (r=0.989; p<0.001).  Table 1. shows the comparisons between the group means and standard deviations for the six categories of trauma.  There were significant differences between the groups for all trauma categories except sexual abuse, which had been reported by only one victim.  The most significant differences were in psychological abuse and overall trauma.
Table 1 about here
 Among the victim group, 8 had undergone extensive psychotherapy, and 6 had completed their treatment at least two years prior to the study.  No subject at the time of the study complained of ongoing psychiatric symptoms, and each was judged, by the interviewing psychiatrist, to have no active DSM IIIR diagnosis on axis I, II, or III.  The 8 subjects who had been in psychotherapy had a past history of a DSM-IIIR diagnosis, including major depression, dysthymia, and substance abuse, but none had ever manifest psychotic symptoms.  No subject in the control group had complained of significant psychological symptoms, and none had been in psychotherapy or had any DSM-IIIR diagnosis in the past.
Table 2 about here
 Immediately after the neutral task, each subject reported that he was able to cooperate in remembering a work or school situation.   There was no significant difference between the POMS scores of the 2 groups during the neutral memories.  During the unpleasant memory on the abbreviated POMS scale, 9 of 10 victims and 5 of 10 controls reported a "quite a bit" or "extreme" response for at least one affect category.  Table 2. shows that subjects in both groups had higher mean abbreviated POMS scores following the unpleasant memory than following the neutral memory.  However, there were no significant differences between groups in abbreviated POMS response to the two memory conditions.  Of the 7 affects measured, the largest change for both groups was for sadness, with victims more affected than controls.  The only POMS items that were differentially affected by group were sadness (p = 0.043), hopelessness (p < 0.001), and mean of the 7 POMS items (p = 0.011).
Figure 1 about here

 There were significant changes in the asymmetry index as a consequence of the neutral vs unpleasant memory condition (F =8.61, df = 1,18, p = 0.009), and there was a significant memory condition by group interaction (F = 6.32, df = 1,18, p = 0.02).  Figure 1 displays representative probe evoked potential tracings from a subject with a trauma history during the neutral and unpleasant memory states, revealing shifts in N1 - P2 amplitude over left and right auditory cortex between the two memory tasks.  Figure 2 shows the average asymmetry index on each task for the control and victim groups.  The victim group displayed a significant left dominant asymmetry during the neutral memory (asymmetry index =-0.159; t = 2.96, df = 9, p < 0.02),  and relative right dominance during the unpleasant memory (asymmetry index = +0.122; t = 1.91, p < 0.10).  Overall, these subjects displayed a highly significant shift in their asymmetry index between memory conditions (t=  3.469, df=9, p=0.007).  During the neutral task, 9 victims had higher amplitudes at C4 than C3 (asymmetry index < 0), implying greater relative left-sided cortical activity.  During the unpleasant memory task 7 victims had lower N1-P2 amplitudes at C4 than C3 (asymmetry index > 0), implying greater relative right-sided cortical activity.
Figure 2 about here
 On the other hand, the control group did not display a significant asymmetry during either the neutral task (asymmetry index = -0.021; t = 0.037, ns), nor the unpleasant memory task (asymmetry index = +0.001; t = 0.02, ns).  Nor was there a significant shift in asymmetry index between the two tasks (t = -0.334, df=9, p = 0.746).  During the neutral task, 60% controls had higher amplitudes at C4 than C3, vs. 40% during the unpleasant memory task.
 Within each group there was no correlation between the severity of overall trauma history and the shift in asymmetry, but for the entire group, the correlation approached significance (r = 0.407, p = 0.075).  Within each group the range of ratings was narrow.  Seven victims had overall abuse ratings of "extreme" or "severe" and 3 had "moderate" ratings.  In the control group all were rated "mild" or "none."
 For the entire group there were no significant correlations between the degree of shift in asymmetry between tasks and degree of emotional response as reflected in average POMS scores (r = -0.058, p > 0.8), or in any individual POMS item.  Also, within groups there was no significant positive correlation between the average change in the POMS scores and the shift in asymmety.
  Analysis of covariance was used as a partial means of testing whether group differences between controls and victims in the degree of asymmetry shift was a consequence of group differences in degree of emotional response to the memories.  Statistically significant group differences in degree of shift persisted even when change in emotional responses were used as covariates (F[1,17]/sadness = 8.02; F[1,17]/hopeless = 5.39; F[1,17]/mean = 13.51;  all pís < 0.04).  Hence, differences between groups in degree of asymmetry shift did not appear to be a direct consequence of differences in expressed emotional response.

 The present study is limited by small sample size, and we agree with Regan (21) that drawing conclusions about cognitive function from evoked potential data should be done with caution.  Still, probe AEPs have been found to be reliable in demonstrating cerebral asymmetries in a number of studies (17,22-25).  For instance, in one study (23) probe AEP's indicated increased right hemispheric activity when subjects were asked to detect the emotional content of speech rather than phonetic aspects. Probe AEP results have been shown to correlate with measures of cerebral blood flow (26).  Still probe AEP's do rely on a few layers of inference and their precision, like all EEG methods of localizing cerebral function, has not been absolutely determined.     Probe AEPs greatly reduce the possible influence of unilateral muscle artifacts, which are know to contaminate the beta spectrum of standard EEGs recorded under emotionally provocative conditions (27).  Auditory probe evoked potentials, recorded over the primary cortical auditory centers, are less influenced by volume conduction which complicates the interpretations of alpha EEG data.  Electrical measures of brain activity have the advantage of being non-invasive, relatively inexpensive, conveniently applied and relatively non interfering with cognitive processes.  Electrical measures, however, lack the capacity of PET scans or newer MRI techniques (28) to provide measures reflecting changes in blood volume, blood flow, or cortical metabolism, and such technologies will be needed to validate our findings.
 We engaged 20 subjects, 10 with a history of psychological trauma, in a brief psychiatric interview and had them recall distressing early memories.  The authors used a non-standardized, subjective initial selection criteria, but the statistical analysis of the blind ratings of all subjects indicated that the two groups were distinct in regard to trauma.  The measurement of trauma, although apparently reliable, could not be tested for validity.  Our trauma group is unique among psychiatric studies since these victims were not presently suffering from any psychiatric symptoms.  This group did have a high prevalence of prior psychiatric illness, in contrast to the non-traumatized sample, and this, in addition to the trauma history, could have possibly contributed to our findings.
 Although both groups demonstrated a shift in laterality toward the right during the interview, only the victim group demonstrated a significant leftward asymmetry during the neutral memory and a significant shift to the right during the unpleasant memory.  The major finding seems to be that of asymmetry regardless of task.  We did not anticipate group differences in the neutral state, and this observation merits further study.
 Although the interviewing psychiatrist had the impression that both groups had similar types of memories, differing mostly in intensity, it is quite possible that the content and character of the memories were distinct for the two groups in subtle ways.  For example, one group may have had more concrete versus abstract memories.
   The correlation between the magnitude of the left-right shift and the estimated severity of the subjectís early abuse almost achieved statistical significance.  Within groups, however, there was no correlation between the level of abuse and AEP shift.  The lack of correlation within each group may be a consequence of the restricted range of ratings as well as the fact that our measure of the degree of trauma could not take into account the individual's constitution, supports, development, or personal meaning.
 The victim group had a significantly stronger average emotional response to the unpleasant memories than controls on a few POMS items, but within groups, and for all subjects, there was no positive correlation between the degree of emotional responses to the unpleasant memory and degree of shift in laterality.  Analysis of covariance suggested that the observed differences between groups were not an artifact of group differences in emotional response.
 Previous reports have suggested that early abuse may possibly be associated with enduring neruobiolobical effects (29-33).  Our findings seem consistent with a recent study by Ito  et al (29), which demonstrated that early childhood abuse was associated with lateralized electrophysiological and neuropsychological abnormalities.   We agree with Muller (34), and hypothesize that early trauma could possibly lead to a lack of integration of left-right hemisphere function, and we further speculate that traumatic memories may be preferentially stored in the right hemisphere.  This hypothesis of deficient hemispheric integration and preferential right-sided storage of traumatic memories, provides an interesting theoretical explanation for memory recollection following trauma, which can be both deficient (constricted or amnestic) and intrusive (35).


1. Harrington A: Medicine, Mind, and the Double Brain.  Princeton, Princeton University Press, 1987
2. Sperry RW: Hemisphere deconnection and unity in conscious awareness.  Amer Psychologist 1968; 23: 823-733
3.  Silberman EK, Weingartener H:  Hemispheric lateralization of functions related to emotion.  Brain and Cognition 1986; 5: 322-353
             4. Safer MA, Leventhal H:  Ear differences in evaluating emotional tone of voice and of verbal content.  J Exper Psych:  Human Percept and Perform 1977, 3: 75-82
             5. Suberi M, McKeever WF:  Differential right hemispheric memory storage of emotional and non-emotional faces.  Neuropsychologia 1977; 5: 757 768
             6. Ross ED, Edmondson JA, Seibert GB, et al:  Acoustic analysis of affective prosody during right-sided Wada test:  a within-subjects verification of the right hemisphere's role in language.  Brain and Language 1988; 33: 128-145
             7. Sackeim HA, Gur RC:  Lateral asymmetry in intensity of emotional expression.  Neuropsychologia 1978; 116: 473-481
             8. Wechsler AF:  The effect of organic brain disease on recall of emotionally charged verses neutral narrative texts.  Neurology 1973; 23: 130-135
             9. Mater DR, Thompson C, Dunn G, Lishman WA:  Memory selectivity and unilateral cerebral dysfunction.  Psychological Med 1986; 1116: 781 788
            10. Schwartz GE, Davidson RJ, Maer F:  Right hemisphere lateralization for emotion in the human brain:  Interactions with cognition.  Science 1975; 190: 286-288
            11. Tucker DM, Roth RS, Arneson BA, et al:  Right hemisphere activation during stress.  Neuropsychologia 1977; 15: 697-700
            12. Ladavas E, Nicoletti R, Umilta C, Rizzolatti G:  Right hemisphere interference during negative affect:  A reaction time study.  Neuropsychologia 22: 479-485, 1984
            13. Ahern GL, Schwartz GE:  Differential lateralization for positive and negative emotion in the human brain:  EEG spectral analysis.  Neuropsychologia 1985; 23: 745-756
            14. Tomarken AJ, Davidson RJ, Henriques JB:  Resting frontal brain asymmetry predicts affective responses to films.  J Personality Soc Psychol 1990; 59: 791-801
            15. Davidson RJ, Ekan P, Saron DC, Senulis J, Friesen W:  Approach/withdrawal and cerebral asymmetry:  Emotional expression and brain physiology I.  J Personality Soc Psychol 1990; 58: 330-341
            16. Tucker DM, Stenslie CE, Roth RS, et al:  Right frontal lobe activation and right hemisphere performance: Decrement during a depressed mood.  Arch Gen Psychiatry 1981; 38: 169-174
            17. Papanicolaou AC, Johnstone J:  Probe evoked potentials: Theory, method and applications.  Intern J Neuroscience 1984; 24: 107-131
            18. Oldfield RC:  The assessment and analysis of handedness:  The Edinburgh Inventory.  Neuropsychologia 1971; 9: 97-113
            19. Lorr M, McNair DM, Droppleman LF:  Manual:  profile of mood states. San Diego, Educational and Industrial Testing Service, 1971
  20. McNair DM, Frankenthaler LM, Czerlinsky T, White TW, Sasson S, Fisher S: Simulated public speaking as a model of clinical anxiety. Psychopharmacology 1982;77:7-10
21. Regan D:  Human Brain Electrophysiology:  Evoked Potentials and Evoked Magnetic Fields in Science and Medicine.  New York, Elsevier, 1989
22. Papanicolaou AC, Schmidt AL, Moore BD, et al:  Cerebral activation patterns in a arithmetic and a visuospatial processing task.  Int J Neurosci 1983; 20: 289-294
             23. Papanicolaou AC, Levin HS, Eisenberg HM, et al:  Evoked potential indices of selective hemispheric engagement in affective and phonetic tasks.  Neuropsychologia 1983; 21: 401-405
             24. Papanicolaou AC, DiScenna A, Gillespie L, et al:  Probe-evoked potential findings following unilateral left-hemisphere lesions in children.  Arch Neurol 1990; 47: 562-566
             25. Armitage R, Bell I, Campbell K, et al:  Asymmetrical auditory probe evoked potentials during REM and NREM sleep.  Sleep 1990; 13: 69-78
             26. Papanicolaou AC, Deutsch G, Bourbon WT, et al:  Convergent evoked potential and cerebral blood flow evidence of task-specific hemispheric differences.  Electroencephalogr Clin Neurophysiol 1987; 66: 512-520
             27. Davidson RJ:  EEG measures of cerebral asymmetry:  Conceptual and methodological issues.  Intern J Neuroscience  1988; 39: 71-89
                      28. Belliveau JW, Kennedy DN, McKinstry RC, et al:  Functional Mapping of the human visual cortex by magnetic resonance imaging.  Science 1991; 254: 716-719
29. Ito Y, Teicher MH, Glod CA, et al: Increased prevalence of electrophysiological abnormalities in children with psychological, physical, and sexual abuse. J Neuropsychiatry Clin Neurosci 1993;5:401-408
30. Teicher MH, Glod CA, Surrey J, et al: Early childhood abuse and limbic system ratings in adult psychiatric outpatients. J Neuropsychiatry Clin Neurosci 1993;5:301-306
31. Green AH, Voeller K, Gaines RW, et al: Neurological impairment in maltreated children. Child Abuse Negl 1981;5:129-134
32. Davies RK: Incest: some neuropsychiatric findings. Int J Psychiatry Med 1979;9:117-121
33. van der Kolk BA: The body keeps the score: memory and the evolving psychobiology of posttraumatic stress. Harvard Rev Psychiatry 1994;1:253-265
34. Muller RJ.  Is there a neural basis for borderline splitting?   Compr. Psychiatry, 1992;33: 92-104
35. van der Kolk BA: The psychological consequences of overwhelming life experiences, in Psychological Trauma, edited by van der Kolk BA. Washington, DC, American Psychiatric Press, 1987, pp 1-30