- Email: [email protected]
Latency of the P3 event-related potential: Normative aspects and within-subject variability
420
Electroencephalography and clinical Neurophysiology, 1984, 59:420-424 Elsevier Scientific Publishers Ireland, Ltd.
LATENCY O F T H E P3 E V E N T - R E L A T E D P O T E N T I A L : N O R M A T I V E A S P E C T S AND W I T H I N S U B J E C T VARIABILITY t DANIEL A. SKLARE 2 and GEORGE E. LYNN
Department of Neurology, and Department of Audiology, Wayne State University School of Medicine, and the Holden Laboratory of Clinical Neurophysiology, Harper-Grace Hospitals, Detroit, M1 48201 (U.S.A.) (Accepted for publication: May 2, 1984)
In recent years, there has been a growing interest in neurophysiological correlates of human information processing. A neurophysiological correlate of selective attention is a late positive scalp-recorded event-related potential that has a latency of approximately 300 msec. This potential, called the P3 or P300 potential, is generated when a subject detects an unexpected but relevant stimulus in the auditory, visual or somatic modalities. A review of the literature reveals no systematic investigation of the stability of the P3 potential latency over time, a prerequisite for valid comparisons between repeated measures. The present investigation examines within-subject test-retest reliability of P3 latency measures obtained from young adults across trials within one test session and across test sessions separated by 2 - 4 weeks. As a corollary, P3 latency norms for the age group and experimental paradigm employed in the investigation were obtained.
Method
Twenty neurologically and audiologically normal young adults (8 male, 12 female) within the age range 22-34 years served as subjects for this investigation. All subjects had hearing thresholds which did not exceed 20 dB H L (ANSI 1969) in
t This study was supported by the Wayne State University School of Medicine Neuroscience Research Program. 2 Reprint requests should be sent to Daniel A. Sklare, Department of Neurology, Wayne State University School of Medicine, 4201 St. Antoine, UHC 6E, Detroit, MI 48201, U.S.A.
both ears at 1 and 2 kHz, with interaural differences within 15 dB HL. Stimuli consisted of 400 tone bursts with a condensation starting phase 20 msec in duration (plateau time) with a 9.9 msec rise/fall time presented binaurally via earphones at an intensity of 88 dB peak sound pressure and at a rate of 1 tone burst every 2 sec. Eighty percent of the 400 tones were 1 kHz (frequent) and 20% were 2 kHz (rare). Stimulus sequence was random. The signals were in phase at the two ears. Electroencephalographic responses were recorded with standard gold disc electrodes located at the vertex and linked earlobes and amplified 8000 times. Evoked responses to the frequent and rare stimuli were filtered with a bandpass 1-30 Hz (filter slope = 12 dB/octave) and averaged simultaneously, but separately over an analysis interval of 768 msec using 256 data points. D a t a from 2 trials were obtained consecutively and stored. Electrode resistance was maintained at 2 kI2 or less. The P3 potential was defined as the first positive wave or wave form complex after 250 msec. Peak latencies of the P3 component for the rare tones were measured by cursor to the nearest millisecond relative to stimulus onset. The peak was defined as the highest point unless the wave was broad or there were multiple peaks. In these latter instances, the peak was defined as the calculated midpoint of the distance measured between the leading and trailing slopes of the wave. All testing was conducted in a standard audiometric sound-treated room during daylight hours. Subjects were in a rested state and had not in-
0168-5597/84/$03.00 © 1984 Elsevier Scientific Publishers Ireland, Ltd.
LATENCY OF P3: WITHIN-SUBJECT VARIABILITY
gested any drugs or significant amounts of caffein prior to the test. Subjects were seated in a comfortable reclining chair and were instructed to disregard the low pitch (1 kHz) tones and to mentally count and keep a record of the high pitch (2 kHz) tones heard and report that total following the trial sequence. In all cases, the reported total was correct within 3 stimuli of the actual total of rare tones presented. Subjects were instructed to relax, avoid movement, and confine their gaze to a circular marking. The visual fixation task has been shown to effectively eliminate electro-oculographic artifact in contingent negative variation (CNV) recordings of cooperative adult subjects (Hillyard and Galambos 1970; Wasman et al. 1970), and has been effectively utilized during conventional longlatency auditory evoked potential recordings (Vaughan and Ritter 1970) as well. Its use is therefore felt justified in the present investigation, although electrophysiologic monitoring of eye movement potentials was not performed. After 2 - 4 weeks, 9 subjects returned for repeat P3 measurement (test session II) involving an identical test and recording protocol.
421
Frequent
Rare P3
Number
PZ
150 175
300
600 msec
800 0
150 17'3 300 350 msec
6(~0
800
Fig. 1. A comparison of evoked potential recordings obtained by the frequent and rare stimuli from 3 subjects during one test session. The time scale is in milliseconds relative to stimulus onset (arrow). The P3 potential is prominent in the rare stimuli recordings with variations in wave form among the 3 subjects. Amplitudes of the frequent and rare recordings reflect 80 and 20% of the total number of stimuli presentations, respectively. Frequent recordings are not normalized. Frequent calibration bar = 2.04 #V, rare calibration bar = 8.17 ~V.
A linear regression analysis demonstrated that the rate of latency increase (slope) was 2.01 msec/year. All but one subject had P3 latencies within 2 S.E. of estimate of the regression line (1 S.E. = 12.52 ITIsec).
Results
The evoked potential wave forms elicited for both the frequent and rare stimuli are shown for 3 subjects in Fig. 1. When subjects selectively attended to the rare stimuli, a large P3 was present in those recordings; this potential was absent or markedly attenuated in the recordings of the frequent stimuli. Responses to frequent stimuli were not normalized. Variations in P3 wave form morphology across subjects were categorized as single-peaked (broad and narrow), bifid-peaked, and multi-peaked, and are shown in Fig. 1. Mean latencies for the 2 trials of test session I ranged from 282.0 to 331.5 msec with an overall mean of 306.5 msec and a standard deviation of 14.7 msec. All mean latencies across subjects (N = 20) fell within 2 S.D. of the overall mean. A Pearson product-moment correlation coefficient (r) of 0.56 indicated a moderate positive correlation between age and P3 latency ( P < 0.05).
A Pearson product-moment correlation coefficient (r) of 0.84 demonstrated a strong positive correlation between latency on trial 1 and trial 2 across subjects ( P < 0.05). A one-factor repeated measures design A N O V A demonstrated that P3 values between trials were significantly different, F (1, 19) = 5.67, P < 0.05, with trial 1 showing longer mean latencies than trial 2 by 4.7 msec. A sub-sample (N = 9) of the experimental group returned for repeat P3 measurement after 2 - 4 weeks (test session II). P3 latency data for this test-retest group are shown in Table I. Within-subject intertest differences in the mean P3 latency range from 1.5 to 12.0 msec, with an overall mean difference of 6.3 msec and a standard deviation of 3.9 msec. A Pearson product-moment correlation coefficient (r) of 0.93 indicated a strong positive correlation between mean latencies of test session I and test session II ( P < 0.05). A two-factor repeated measures A N O V A did not demonstrate statistically significant differences in latencies as a
422
D.A. SKLARE, G.E. LYNN
TABLE I P3 test-retest latency (msec) values (N = 9). Subject no.
Age (years)
Test session I
Difference
Trial 1
Trial 2
-XI
Trial 1
Trial 2
X,n
1 2 3 4 5 6 7 8 9
33 24 31 30 23 32 26 23 22
324.0 291.0 331.5 304.5 303.0 312.0 333.0 297.0 288.0
321.0 291.0 317.5 291.0 294.0 312.0 330.0 285.0 297.0
322.5 291.0 324.5 297.8 298.5 312.0 331.5 291.0 292.5
312.0 285.0 321.0 288.0 291.0 312.0 336.0 300.0 294.0
309.0 288.0 309.0 285.0 294.0 309.0 318.0 294.0 294.0
310.5 286.5 315.0 286.5 292.5 310.5 327.0 297.0 294.0
12.0 4.5 9.5 11.3 6.0 1.5 4.5 - 6.0 - 1.5
S.D.
27.1 4.37
309.3 16.9
304.3 16.0
306.8 16.0
304.3 17.1
300.0 11.4
302.2 14.1
6.3 3.9
Test session II
function of either tests or trials ( P < 0.05). Furthermore, the interaction between tests and trials was not significant.
Discussion
Normative data collected during the present study reflect the same general trends of positive correlation of P3 latency with aging and intersubject variability within 2 S.D. demonstrated in previous reports (Marsh and Thompson 1972; Brent et al. 1976; Goodin et al. 1978a,b; Syndulko et al. 1982). This age-latency relationship was seen even within the restricted age range of 12 years (22-34). However, the actual value of the correlation coefficient was smaller and the slope was larger in this study compared to Goodin et al. (1978a) and Syndulko et al. (1982). This might have been due to the larger sample size within a very restricted age range used in the present study. Results of the present study are felt to support the hypothesis that the P3 potential is a reliable measure, having the stability required for clinical and research applications. Although a statistically significant difference between trials within a test session was demonstrated for 20 subjects, the small mean latency difference between trials (4.7 msec) is not felt to be of practical importance because of the high degree of intersubject variability in P3
X I -X n
latency and the high correlation in the subjects' performance between trials. It is possible that practice effects or heightened arousal to mitigate against fatigue during the second trial may account for the slightly shorter latencies found in 13 subjects and on the average during the second trial. Statistical analysis of the test-retest group (N = 9) did not reveal significant latency differences between trials and tests owing to a reduction in the power of the ANOVA with a smaller sample size. The P3 potential wave form was either singlepeaked, bifid-peaked or multi-peaked. This variation in P3 wave form morphology has been interpreted as the manifestion of two distinct P3 subcomponents as identified by N.K. Squires et al. (1975) and K.C. Squires et al. (1977). These subcomponents, termed P3a and P3b have been shown to differ in their latency, scalp topography and psychological correlates (N.K. Squires et al. 1975). Variability in P3 wave form morphology may therefore reflect the degree to which P3 sub-components fuse or separate. As only a central recording site was utilized in the present investigation, separation of the P3 sub-components (N.K. Squires et al. 1975) was not possible in this study. However, in clinical applications where the P3 complex is often poorly defined, multiple recording sites may improve identification of P3 and thus enhance the clinical utility of P3 studies. Since eyeblinks and eye movements introduce
LATENCY OF P3: WITHIN-SUBJECT VARIABILITY considerable artifact to the neuroelectric response (Hillyard and G a l a m b o s 1970; Girton and K a m i y a 1973), monitoring of ocular potentials is particularly important in clinical studies, as m a n y patients cannot maintain visual fixation during P3 recordings. The test-retest reliability of P3 latency found in this study over a period of 2 - 4 weeks substantiates its usefulness to the study of normal and disordered h u m a n cognitive processes in y o u n g adults. The stability of P3 latency in older subjects should be investigated to determine the clinical utility of the P3 potential across the life span.
Summary A growing b o d y of research has focused on the P3 (P300) event-related potential as an electrophysiological correlate of selective attention. The present investigation examines the intrasubject test-retest reliability of the auditory evoked P3 latency measure for neurologically and audiologically normal y o u n g adults aged 2 2 - 3 4 years across test sessions separated by 2 - 4 weeks (N = 9) and across trials within one test session (N = 20). In a target-selection ( ' o d d b a l l ' ) paradigm, subjects mentally c o u n t e d infrequent 2 k H z tone bursts (targets) r a n d o m l y interspersed within a sequence of 1 k H z tone bursts (non-targets). A strong positive correlation was demonstrated between latencies of test sessions I and II. A n analysis of variance did not demonstrate statistically significant latency differences as a function of either tests or trials for the test-retest group (N = 9). A l t h o u g h analysis of variance demonstrated a statistically significant difference between trials within one test session for 20 subjects, the small mean latency difference between trials (4.7 msec) is interpreted as being clinically unimportant. The stability of P3 latency found in this study over a period of 2 - 4 weeks supports its application to the study of normal and disordered cognitive processes.
423
R~sum6 Latence de l'onde P3 du potentiel lib ~t l'bvbnement: aspects normatifs et variabilitb pour un m~me sujet De plus en plus d'6tudes se concentrent sur l'onde P3 (P300) du potentiel li6 h l'6v6nement, c o m m e 616ment 61ectrophysiologique 1i6 h l'attention s61ective. Le pr6sent travail analyse chez de jeunes adultes de 2 2 - 3 4 ans, neurologiquement et audiologiquement normaux, la fiabilit6 de la mesure de la latence de P3 de la r6ponse 6voqu6e auditive chez un m~me sujet, ceci d ' u n test h rautre, soit entre sessions espac6es de 2 h 4 semaines (N = 9), soit entre essais au cours de la m~me session (N = 20). Dans un paradigme de s61ection de cible ('oddball'), les sujets comptaient mentalement des bouff6es tonales peu fr6quentes de 2 k H z (cible), dispers6es au hasard dans une s6quence de bouff6es tonales de 1 k H z (non-cible). U n e forte corr61ation positive a 6t6 mise en 6vidence entre les latences des sessions de test I et II. U n e analyse de la variance n'a pas montr6 de diff6rence de latence statistiquement significatives en fonction des s6ries de test ou des essais dans le groupe test6 plusieurs fois (N = 9). Bien que l'analyse de variance ait d6montr6 pour 20 sujets, une diff6rence statistiquement significative entre les essais au cours d ' u n e m~me session, la faible diff6rence entre les moyennes des latences entre essais (4,7 msec) a 6t6 interpr6t6e c o m m e sans utilit6 clinique. La stabilit6 de la latence de P3 mise en 6vidence dans cette 6tude sur des p6riodes de 2 - 4 semaines encourage l'application de cette mesure dans l'6tude des processus cognitifs n o r m a u x et anormaux. We would like to acknowledge the assistance of Dr. Doris V. Allen who aided in the statistical analysis of this study and commented on the manuscript, and to Dr. George Corondan for the R6sum6.
References Brent, G.A., Smith, D.B.D., Michalweski, H.J. and Thompson, L.W. Differences in the evoked potential in young and old subjects during the habituation and dishabituation procedures. Psychophysiology, 1976, 14: 96-97. Girton, D.G. and Kamiya, J. Simple on-line technique for
424
removing eye movement artifacts from the EEG. Electroenceph. clin. Neurophysiol., 1973, 34: 212-216. Goodin, D.S., Squires, K.C., Henderson, B.H. and Starr, A. Age-related variations in evoked potentials to auditory stimuli in normal human subjects. Electroenceph. clin. Neurophysiol., 1978a, 44: 447-558. Goodin, D.S., Squires, K.C. and Starr, A. Long latency eventrelated components of the auditory evoked potential in dementia. Brain, 1978b. 101: 635-648. Hillyard, S.A. and Galambos, R. Eye movement artifact in the CNV. Electroenceph. clin. Neurophysiol., 1970, 28: 173-182. Marsh, G. and Thompson, E. Age differences in evoked potentials during an auditory discrimination task. Gerontologist, 1972, 12: 44. Squires, K.C., Donchin, E., Herning. R.I. and McCarthy, G. On the influence of task relevance and stimulus probability
D.A. SKLARE,
G.E. LYNN
on event-related-potential components. Electroenceph. clin. Neurophysiol., 1977, 42: I-14. Squires, N.K., Squires, K.C. and Hillyard, S.A. Two varieties of long-latency positive waves evoked by unpredictable auditory stimuli in man. Electroenceph. clin. Neurophysiol., 1975, 38: 387-401. Syndulko, K., Hansch, E.C., Cohen, S.N., Pearce, J.W.. Goldberg, Z., Montan, B., Tourtellotte, W.W. and Potvin, A.R. Long-latency event-related potentials in normal aging and dementia. In: J. Courjon, F. Mauguiere and M. Revel (Eds.), Clinical Applications of Evoked Potentials in Neurology. Raven Press, New York, 1982: 279-285. Vaughan, H.G. and Ritter, W. The source of auditory evoked responses recorded from the human scalp. Electroenceph. clin. Neurophysiol., 1970, 28: 360-367. Wasman, M., Morehead, S.D., Lee, H. and Rowland, J. Interaction of electro-ocular potentials with the contingent negative variation. Psychophysiology, 1970, 1: 1033111.
Electroencephalography and clinical Neurophysiology, 1984, 59:420-424 Elsevier Scientific Publishers Ireland, Ltd.
LATENCY O F T H E P3 E V E N T - R E L A T E D P O T E N T I A L : N O R M A T I V E A S P E C T S AND W I T H I N S U B J E C T VARIABILITY t DANIEL A. SKLARE 2 and GEORGE E. LYNN
Department of Neurology, and Department of Audiology, Wayne State University School of Medicine, and the Holden Laboratory of Clinical Neurophysiology, Harper-Grace Hospitals, Detroit, M1 48201 (U.S.A.) (Accepted for publication: May 2, 1984)
In recent years, there has been a growing interest in neurophysiological correlates of human information processing. A neurophysiological correlate of selective attention is a late positive scalp-recorded event-related potential that has a latency of approximately 300 msec. This potential, called the P3 or P300 potential, is generated when a subject detects an unexpected but relevant stimulus in the auditory, visual or somatic modalities. A review of the literature reveals no systematic investigation of the stability of the P3 potential latency over time, a prerequisite for valid comparisons between repeated measures. The present investigation examines within-subject test-retest reliability of P3 latency measures obtained from young adults across trials within one test session and across test sessions separated by 2 - 4 weeks. As a corollary, P3 latency norms for the age group and experimental paradigm employed in the investigation were obtained.
Method
Twenty neurologically and audiologically normal young adults (8 male, 12 female) within the age range 22-34 years served as subjects for this investigation. All subjects had hearing thresholds which did not exceed 20 dB H L (ANSI 1969) in
t This study was supported by the Wayne State University School of Medicine Neuroscience Research Program. 2 Reprint requests should be sent to Daniel A. Sklare, Department of Neurology, Wayne State University School of Medicine, 4201 St. Antoine, UHC 6E, Detroit, MI 48201, U.S.A.
both ears at 1 and 2 kHz, with interaural differences within 15 dB HL. Stimuli consisted of 400 tone bursts with a condensation starting phase 20 msec in duration (plateau time) with a 9.9 msec rise/fall time presented binaurally via earphones at an intensity of 88 dB peak sound pressure and at a rate of 1 tone burst every 2 sec. Eighty percent of the 400 tones were 1 kHz (frequent) and 20% were 2 kHz (rare). Stimulus sequence was random. The signals were in phase at the two ears. Electroencephalographic responses were recorded with standard gold disc electrodes located at the vertex and linked earlobes and amplified 8000 times. Evoked responses to the frequent and rare stimuli were filtered with a bandpass 1-30 Hz (filter slope = 12 dB/octave) and averaged simultaneously, but separately over an analysis interval of 768 msec using 256 data points. D a t a from 2 trials were obtained consecutively and stored. Electrode resistance was maintained at 2 kI2 or less. The P3 potential was defined as the first positive wave or wave form complex after 250 msec. Peak latencies of the P3 component for the rare tones were measured by cursor to the nearest millisecond relative to stimulus onset. The peak was defined as the highest point unless the wave was broad or there were multiple peaks. In these latter instances, the peak was defined as the calculated midpoint of the distance measured between the leading and trailing slopes of the wave. All testing was conducted in a standard audiometric sound-treated room during daylight hours. Subjects were in a rested state and had not in-
0168-5597/84/$03.00 © 1984 Elsevier Scientific Publishers Ireland, Ltd.
LATENCY OF P3: WITHIN-SUBJECT VARIABILITY
gested any drugs or significant amounts of caffein prior to the test. Subjects were seated in a comfortable reclining chair and were instructed to disregard the low pitch (1 kHz) tones and to mentally count and keep a record of the high pitch (2 kHz) tones heard and report that total following the trial sequence. In all cases, the reported total was correct within 3 stimuli of the actual total of rare tones presented. Subjects were instructed to relax, avoid movement, and confine their gaze to a circular marking. The visual fixation task has been shown to effectively eliminate electro-oculographic artifact in contingent negative variation (CNV) recordings of cooperative adult subjects (Hillyard and Galambos 1970; Wasman et al. 1970), and has been effectively utilized during conventional longlatency auditory evoked potential recordings (Vaughan and Ritter 1970) as well. Its use is therefore felt justified in the present investigation, although electrophysiologic monitoring of eye movement potentials was not performed. After 2 - 4 weeks, 9 subjects returned for repeat P3 measurement (test session II) involving an identical test and recording protocol.
421
Frequent
Rare P3
Number
PZ
150 175
300
600 msec
800 0
150 17'3 300 350 msec
6(~0
800
Fig. 1. A comparison of evoked potential recordings obtained by the frequent and rare stimuli from 3 subjects during one test session. The time scale is in milliseconds relative to stimulus onset (arrow). The P3 potential is prominent in the rare stimuli recordings with variations in wave form among the 3 subjects. Amplitudes of the frequent and rare recordings reflect 80 and 20% of the total number of stimuli presentations, respectively. Frequent recordings are not normalized. Frequent calibration bar = 2.04 #V, rare calibration bar = 8.17 ~V.
A linear regression analysis demonstrated that the rate of latency increase (slope) was 2.01 msec/year. All but one subject had P3 latencies within 2 S.E. of estimate of the regression line (1 S.E. = 12.52 ITIsec).
Results
The evoked potential wave forms elicited for both the frequent and rare stimuli are shown for 3 subjects in Fig. 1. When subjects selectively attended to the rare stimuli, a large P3 was present in those recordings; this potential was absent or markedly attenuated in the recordings of the frequent stimuli. Responses to frequent stimuli were not normalized. Variations in P3 wave form morphology across subjects were categorized as single-peaked (broad and narrow), bifid-peaked, and multi-peaked, and are shown in Fig. 1. Mean latencies for the 2 trials of test session I ranged from 282.0 to 331.5 msec with an overall mean of 306.5 msec and a standard deviation of 14.7 msec. All mean latencies across subjects (N = 20) fell within 2 S.D. of the overall mean. A Pearson product-moment correlation coefficient (r) of 0.56 indicated a moderate positive correlation between age and P3 latency ( P < 0.05).
A Pearson product-moment correlation coefficient (r) of 0.84 demonstrated a strong positive correlation between latency on trial 1 and trial 2 across subjects ( P < 0.05). A one-factor repeated measures design A N O V A demonstrated that P3 values between trials were significantly different, F (1, 19) = 5.67, P < 0.05, with trial 1 showing longer mean latencies than trial 2 by 4.7 msec. A sub-sample (N = 9) of the experimental group returned for repeat P3 measurement after 2 - 4 weeks (test session II). P3 latency data for this test-retest group are shown in Table I. Within-subject intertest differences in the mean P3 latency range from 1.5 to 12.0 msec, with an overall mean difference of 6.3 msec and a standard deviation of 3.9 msec. A Pearson product-moment correlation coefficient (r) of 0.93 indicated a strong positive correlation between mean latencies of test session I and test session II ( P < 0.05). A two-factor repeated measures A N O V A did not demonstrate statistically significant differences in latencies as a
422
D.A. SKLARE, G.E. LYNN
TABLE I P3 test-retest latency (msec) values (N = 9). Subject no.
Age (years)
Test session I
Difference
Trial 1
Trial 2
-XI
Trial 1
Trial 2
X,n
1 2 3 4 5 6 7 8 9
33 24 31 30 23 32 26 23 22
324.0 291.0 331.5 304.5 303.0 312.0 333.0 297.0 288.0
321.0 291.0 317.5 291.0 294.0 312.0 330.0 285.0 297.0
322.5 291.0 324.5 297.8 298.5 312.0 331.5 291.0 292.5
312.0 285.0 321.0 288.0 291.0 312.0 336.0 300.0 294.0
309.0 288.0 309.0 285.0 294.0 309.0 318.0 294.0 294.0
310.5 286.5 315.0 286.5 292.5 310.5 327.0 297.0 294.0
12.0 4.5 9.5 11.3 6.0 1.5 4.5 - 6.0 - 1.5
S.D.
27.1 4.37
309.3 16.9
304.3 16.0
306.8 16.0
304.3 17.1
300.0 11.4
302.2 14.1
6.3 3.9
Test session II
function of either tests or trials ( P < 0.05). Furthermore, the interaction between tests and trials was not significant.
Discussion
Normative data collected during the present study reflect the same general trends of positive correlation of P3 latency with aging and intersubject variability within 2 S.D. demonstrated in previous reports (Marsh and Thompson 1972; Brent et al. 1976; Goodin et al. 1978a,b; Syndulko et al. 1982). This age-latency relationship was seen even within the restricted age range of 12 years (22-34). However, the actual value of the correlation coefficient was smaller and the slope was larger in this study compared to Goodin et al. (1978a) and Syndulko et al. (1982). This might have been due to the larger sample size within a very restricted age range used in the present study. Results of the present study are felt to support the hypothesis that the P3 potential is a reliable measure, having the stability required for clinical and research applications. Although a statistically significant difference between trials within a test session was demonstrated for 20 subjects, the small mean latency difference between trials (4.7 msec) is not felt to be of practical importance because of the high degree of intersubject variability in P3
X I -X n
latency and the high correlation in the subjects' performance between trials. It is possible that practice effects or heightened arousal to mitigate against fatigue during the second trial may account for the slightly shorter latencies found in 13 subjects and on the average during the second trial. Statistical analysis of the test-retest group (N = 9) did not reveal significant latency differences between trials and tests owing to a reduction in the power of the ANOVA with a smaller sample size. The P3 potential wave form was either singlepeaked, bifid-peaked or multi-peaked. This variation in P3 wave form morphology has been interpreted as the manifestion of two distinct P3 subcomponents as identified by N.K. Squires et al. (1975) and K.C. Squires et al. (1977). These subcomponents, termed P3a and P3b have been shown to differ in their latency, scalp topography and psychological correlates (N.K. Squires et al. 1975). Variability in P3 wave form morphology may therefore reflect the degree to which P3 sub-components fuse or separate. As only a central recording site was utilized in the present investigation, separation of the P3 sub-components (N.K. Squires et al. 1975) was not possible in this study. However, in clinical applications where the P3 complex is often poorly defined, multiple recording sites may improve identification of P3 and thus enhance the clinical utility of P3 studies. Since eyeblinks and eye movements introduce
LATENCY OF P3: WITHIN-SUBJECT VARIABILITY considerable artifact to the neuroelectric response (Hillyard and G a l a m b o s 1970; Girton and K a m i y a 1973), monitoring of ocular potentials is particularly important in clinical studies, as m a n y patients cannot maintain visual fixation during P3 recordings. The test-retest reliability of P3 latency found in this study over a period of 2 - 4 weeks substantiates its usefulness to the study of normal and disordered h u m a n cognitive processes in y o u n g adults. The stability of P3 latency in older subjects should be investigated to determine the clinical utility of the P3 potential across the life span.
Summary A growing b o d y of research has focused on the P3 (P300) event-related potential as an electrophysiological correlate of selective attention. The present investigation examines the intrasubject test-retest reliability of the auditory evoked P3 latency measure for neurologically and audiologically normal y o u n g adults aged 2 2 - 3 4 years across test sessions separated by 2 - 4 weeks (N = 9) and across trials within one test session (N = 20). In a target-selection ( ' o d d b a l l ' ) paradigm, subjects mentally c o u n t e d infrequent 2 k H z tone bursts (targets) r a n d o m l y interspersed within a sequence of 1 k H z tone bursts (non-targets). A strong positive correlation was demonstrated between latencies of test sessions I and II. A n analysis of variance did not demonstrate statistically significant latency differences as a function of either tests or trials for the test-retest group (N = 9). A l t h o u g h analysis of variance demonstrated a statistically significant difference between trials within one test session for 20 subjects, the small mean latency difference between trials (4.7 msec) is interpreted as being clinically unimportant. The stability of P3 latency found in this study over a period of 2 - 4 weeks supports its application to the study of normal and disordered cognitive processes.
423
R~sum6 Latence de l'onde P3 du potentiel lib ~t l'bvbnement: aspects normatifs et variabilitb pour un m~me sujet De plus en plus d'6tudes se concentrent sur l'onde P3 (P300) du potentiel li6 h l'6v6nement, c o m m e 616ment 61ectrophysiologique 1i6 h l'attention s61ective. Le pr6sent travail analyse chez de jeunes adultes de 2 2 - 3 4 ans, neurologiquement et audiologiquement normaux, la fiabilit6 de la mesure de la latence de P3 de la r6ponse 6voqu6e auditive chez un m~me sujet, ceci d ' u n test h rautre, soit entre sessions espac6es de 2 h 4 semaines (N = 9), soit entre essais au cours de la m~me session (N = 20). Dans un paradigme de s61ection de cible ('oddball'), les sujets comptaient mentalement des bouff6es tonales peu fr6quentes de 2 k H z (cible), dispers6es au hasard dans une s6quence de bouff6es tonales de 1 k H z (non-cible). U n e forte corr61ation positive a 6t6 mise en 6vidence entre les latences des sessions de test I et II. U n e analyse de la variance n'a pas montr6 de diff6rence de latence statistiquement significatives en fonction des s6ries de test ou des essais dans le groupe test6 plusieurs fois (N = 9). Bien que l'analyse de variance ait d6montr6 pour 20 sujets, une diff6rence statistiquement significative entre les essais au cours d ' u n e m~me session, la faible diff6rence entre les moyennes des latences entre essais (4,7 msec) a 6t6 interpr6t6e c o m m e sans utilit6 clinique. La stabilit6 de la latence de P3 mise en 6vidence dans cette 6tude sur des p6riodes de 2 - 4 semaines encourage l'application de cette mesure dans l'6tude des processus cognitifs n o r m a u x et anormaux. We would like to acknowledge the assistance of Dr. Doris V. Allen who aided in the statistical analysis of this study and commented on the manuscript, and to Dr. George Corondan for the R6sum6.
References Brent, G.A., Smith, D.B.D., Michalweski, H.J. and Thompson, L.W. Differences in the evoked potential in young and old subjects during the habituation and dishabituation procedures. Psychophysiology, 1976, 14: 96-97. Girton, D.G. and Kamiya, J. Simple on-line technique for
424
removing eye movement artifacts from the EEG. Electroenceph. clin. Neurophysiol., 1973, 34: 212-216. Goodin, D.S., Squires, K.C., Henderson, B.H. and Starr, A. Age-related variations in evoked potentials to auditory stimuli in normal human subjects. Electroenceph. clin. Neurophysiol., 1978a, 44: 447-558. Goodin, D.S., Squires, K.C. and Starr, A. Long latency eventrelated components of the auditory evoked potential in dementia. Brain, 1978b. 101: 635-648. Hillyard, S.A. and Galambos, R. Eye movement artifact in the CNV. Electroenceph. clin. Neurophysiol., 1970, 28: 173-182. Marsh, G. and Thompson, E. Age differences in evoked potentials during an auditory discrimination task. Gerontologist, 1972, 12: 44. Squires, K.C., Donchin, E., Herning. R.I. and McCarthy, G. On the influence of task relevance and stimulus probability
D.A. SKLARE,
G.E. LYNN
on event-related-potential components. Electroenceph. clin. Neurophysiol., 1977, 42: I-14. Squires, N.K., Squires, K.C. and Hillyard, S.A. Two varieties of long-latency positive waves evoked by unpredictable auditory stimuli in man. Electroenceph. clin. Neurophysiol., 1975, 38: 387-401. Syndulko, K., Hansch, E.C., Cohen, S.N., Pearce, J.W.. Goldberg, Z., Montan, B., Tourtellotte, W.W. and Potvin, A.R. Long-latency event-related potentials in normal aging and dementia. In: J. Courjon, F. Mauguiere and M. Revel (Eds.), Clinical Applications of Evoked Potentials in Neurology. Raven Press, New York, 1982: 279-285. Vaughan, H.G. and Ritter, W. The source of auditory evoked responses recorded from the human scalp. Electroenceph. clin. Neurophysiol., 1970, 28: 360-367. Wasman, M., Morehead, S.D., Lee, H. and Rowland, J. Interaction of electro-ocular potentials with the contingent negative variation. Psychophysiology, 1970, 1: 1033111.