International Journal of Psychophysiology, 13(1992)233-239 0 1992Elsevier Science Publishers B.V. AI1 rights reserved 0167~8760/92/$05.00
233
INTPSY 00414
e
an
in a
ichi Kameyama b, Shin-Ichi Niwa ‘, Kenji Itoh d, asato Fukuda ‘, Qsamu Saitoh e, Akira Iwanami f, Nakagome g and Tsukasa Sasaki g * Research Institute of Medical Engineering, Faculty of Medicine, Uniuersity of Tokyo, Tokyo (Japan:, b Pasts and Telecommukations Tokyo Hospital, Tokyo (Japan), ’ Department of Neuropsychiatry, Fatuity of Medicine, Unitiersityof Tokyo, Tokyo (Japan), d Research Institute of Logopedics and Phoniatrics, Faculty of Medicine, University of Tokyo, Tokyo (Japan), ’ National Musashi Hospital, Kodaira (Japan), f Tokyo Metropolitan Matsuzawa Hospital, Tokyo (Japan) and s Department of Psychiatry, Faculty of Medicine, Teikyo Utkersity, Tokyo (Japan, (Accepted 28 July 1992)
Key words: Event-Related
Potential; Nd; P300, Normal value; Distribution pattern
TO obtain objective criteria for assessing the attentional and cognitive functioning of psychiatric populations, we attempted to standardize values of two components in Event-Related Potentials (ERPs), namely the attention-related negative potential (Nd) and the P300, in normal populations. The study consisted of 100 healthy volunteers (50 females, 50 males) who were given the task of making dichotic syllable discriminations requiring key-press responses. Their ages ranged between 18 and 59 years (mean f S.D., 32.3 + 11.3 years). Nd was found to be maximum in the Fz region, P300 being maximum in the Pz region. The means and standard deviations of Nd and P300 areas in their maximum regions were 554.1 + 307.8 PV ms and 2148.5 + 1248.5 /.LVms, respectively. The transformation plot for symmetry indicated the suitable power of transformation to be l/2 for both Nd and P300 distributions. After being transformed into square-root values, the distribution patterns of Nd and P300 areas were examined. When the lower limit of normal values was tentatively assigned to mean - 2 S.D. using square-root transformed data for both Nd and P300,97% of the subjects were found to display values above the lower normal limit for Nd, and 98% for the P300. Neither, Nd nor P3OOareas correlated with age, while P300 latencies displayed a w lk positive correlation with age. Females displayed relatively larger values than males for Nd and P300 areas and P300-peak amplituues. However, the differences between females and males were not statistically significant. Females and males showed nearly equal P300-peak latencies.
INTRODUCTION To obtain objective criteria for assessing attentional and cognitive functioning of psychiatric populations, we attempted to standardize values of two components in Event-Related Potentials (ERPs), namely the attention-related negative potential (Nd) and the P300, in normal populations. The Nd refers to negative components of
Correspondence to: S.-I. Niwa, Department of Neuropsychiatry, Faculty of Medicine, University of Tokyo, 7-3-l Hongo, Bunkyo-ku, Tokyo, 113 Japan.
ERPs as have been reported by Naatanen et al. (198, 1981) and Hillyard et al. (1973, 19781, which start at approx. 70-100 ms, peaking at approx. 200 ms after stimulus-onset. They are thought to emerge on occasions of differential discrimination between task-relevant and taskirrelevant stimuli, thus implicating selective attentional functioning. The P300 is a complex of positive components of ERPs that are elicited by novel or rare or task-relevant stimuli at approximately 300 ms after stimulus-onset. Sutton et al. (1965) first explained the possible cognitive meaning of P300 as the resolution of uncertainties. Later, Donchin et
234
al. (1984) hypothesized that the P300 would reflect context-updating or situation-evaluation. The P300 has been regarded as a significant complex reflecting higher cognitive functions. Though we are aware that ERPs include several components other than Nd and P300, we tried to look at these two components because they are steadily and relatively easily recorded and they bear widelyaccepted psychological meanings.
SUBJECTS AND METHODS Subjects 100 healthy adult volunteers (50 male and 50 female; ages 18-59, mean 32.3 f 11.3 years) participated in this study. 94 of the subjects were right handed, 5 were mixed handed, and the remaining one subject was left handed. All of the subjects were free from any hearing disability, and had no history of neurological or psychiatric disorders. The mean of their number of years of education was 14.4 f 2.3 (range 8-22) years. The subjects consisted of 30 university students, 16 engineers, 16 laborers, 10 office workers, 6 doctors, 7 nursing staffs and 15 housewives. There were no differences between female and male subjects in their ages and years of education. Methods In the ~ic~otic syllable discrimination tasks empiuyed m the present study, male and female voices presented the syllables (/te/ and /ga/> to the subjects via heplphones, with the male voice being presented to one ear and the female voice to the other ear with no changes between the stimuli within a session. The frequency-ratio in appearance of /te/ and /ga/ was 1:3. The stimuli were applied in a random, alternative pattern to one side or the other and to only one ear at a time during one session. Subjects were instructed to press a designated response-key to the target with the index or middle finger of the attended side hand, while pressing another key to the other stimuli with the other side hand. In the instruction, both accuracy and speed of responses were equally required. The duration of each stimulus was 150 ms; the stimulus intensity being set
approx. at 60 dBSL. The inter-stimulus intervals -2000 ms. AII of varied randomly between 1 the subjects performed 4 runs, that is, one target syllable X two voices (male or female) X two attended channels (left or right ear). The number of the total stimuli in both ears for each run was approx. 240. The subjects were seated in a sound-proof anechoic room with eyes closed. According to the International lo-20 Electrode System, EEGs were derived from the Fz, Cz and Pz regions monopolarly referenced to linked ear-lobe electrodes. Eye-movements were also recorded bipolarly between above and lateral electrodes to the right eye. EEGs were amplified with bandpass down 6 dB at 0.15 and 300 Hz. After processed through an anti-aliasing filter of approximately 95 Hz, EEGs were digitized on-line by micro PDP 1 l/23 with a sampling frequency of 250 Hz/Ch. EEG data with amplitudes exceeding 100 PV measured between the most negative peak and the most positive through or accompanied by EOGs of more than 150 PV during the period 40 ms prestimulus to 800 ms post-stimulus were eliminated from averaging. In the population employed in the present study, the actual number of rejected EEG responses from averaging remained small. After a smoothing process using a digital filter with a window-width of 32 ms (8 data points), the averaged ERP waveforms for the individual subjects were averaged separately through VAX 8530 into four catories: (1) target syllables in the attended ear; (2) non-target syllables in the attended ear; (3) target syllables in the non-attended ear; and (4) non-target syllables in the nonattended ear. The averaging times for /te/ and /ga/ in one ear were limited to 16 and 48, respectively, for each run, hence producing 64 times for categories 1 and 3 and 192 times for categories 2 and 4 in total, being multiplying by 2 voices X 2 sides. Specifically, first 16 responses for categories 1 and 3 and first 48 responses for 2 and 4 were averaged for each run. It is empirically reasonable to assume that psychiatric patients tend to produce fewer numbers of employable EEG responses for averaging as compared to healthy subjects. The reason of the restriction in averaging times mentioned above is this empir-
ical assumption that requires a balanced averaging times between healthy subjects and pschiatric populations. The Nd was defined as the negative component of a difference wave, in the period of O-400 ms after the stimulus onset, between ERPs ited by non-target syllables in the attended ear and those for non-target syllables in the nonattended ear. For the index to represent Nd, those areas were employed which were defined as the negative areas between O-400 ms in the difference wave described above. P300 was defined as the positive component of ERP, in the period of 260-m ms after the stimulus onset. P300 areas, P300 amplitudes and P300 latencies for targets in the attended ear were employed as indices representing P300. P300 areas were defined as the positive areas of target ERPs between 260-600 ms, with P300 amplitudes and P300 latencies corresponding to those measures of the most positive peaks in the relevant latency periods. Amplitudes were measured with reference to the 0 level defined as the mean amplitude during the 49 ms period just prior to the stimulus onset. RESULTS ERP wacefmms
The grand average of ERP waveforms for one hundred subjects are shown in Fig. 1. Waveforms for the four stimulus categories are shown separately in the figure along with the wave differences. The P3OOsin Fig. 1 are most clearly elicited by the targets in the attended ear particularly at the Pz region. Fig. 1 also reveals that Nd waves are most prominent at the Fz region in which two peaks are identified. In the Cs and Pz regions, the first peak, Cz, is clearer than the second one. The Nd for the Fz region and P300 for the Pz region are utilized in the following analyses. Perforrmxc
and ERPs
Reaction tire-:s and hit rates for the targets were measured in 91 subjects (46 males, 45 females). Indices for the remaining 9 subjects were not measured due to failures in the experimental apparatus. The mean and S.D. of ages for the 91
-1OjfV
k
0
*
-
- 4c(V 0
* t-Q--
-_.__;:-
-
23.5
topv-
?
_.__..- Oi
___---
:.*-r“.., . ‘.. ‘C’_--..*
*
J’-..‘._
‘._.--
*-
__
_A-.
Fz ......... cz --- h
_-
I I I I I I
0
sxhl!T
Fig. 1. The waveforms for the four stimulus categories are separately shown along with the difference waves. Upper left, attend non-target; upper right, attend target; middle left. non-attend non-target; lower right, non-attend target; lower left, difference wave. FL Cz and Pz denote the electrode locations.
subjects was 32.8 f 11.5 years, with these values almost similar to those of the whole subjects. Hit rates for the targets ranged between 85.8 to 100% (mean f S-D., 97.4 f 2.9), with the mean reaction times for individual subjects distributing between 322 to 844 ms (mean + S.D., 545 f 87 ms). Possible correlations between hit rates (or reaction times) and idd (or P300) were examined through calculation of partial correlation coefficients controlling for ages, which revealed the following results: (1) hit rates showed no correlation with the Nd area, P300 area, P300 amplitude and P300 latency. (2) reaction times displayed no correlation with the Nd area, while they demonstrated weak !-.:_:a significant correlations with the P300 area (r = -0.37, P < O.OOl), P300 latency (r = 0.24, P < 0.05) and P300 amplitude (r = - 0.42, P < 0.001). Relationship between the Nd and P300
The means and S.D. of the Nd area at Fz and P300 at Pz were 554.1 f 307.8 PV ms and 2148.5 + 1261.5 PV ms, respectively. The relationship between Nd and P300 areas was not significant (Pearson’s correlation coefficient r = 0.07, n.s.1. The distribution of Nd areas
Since the distributira pattern of absolute Nd area values was asymmetric, with the number of
236
smaller Nd subjects exceeding that of larger Nd subjects (skewness = 0.296), the suitable power of transformation was obtained according to the method of the transformation plot for symmetry (Emerson and Stoto, 1983). This method indicated the power to be l/2. The distribution pattern of square-root transformed Nd values was obtained, of which histogram is displayed in Fig. owever, this distribution reached statistical significance not to follow the Gaussian distribution (X2(5) = 14.24, 0.01 < P < 0.025). The mean and SD. of square-root transformed Nd areas were 22.4 (502.7 PV ms) and 7.1 (50.4 PV ms), respectively. When the lower limit of normal Nd was tentatively defined as mean -2 SD. in square-root transformed Nd areas, it was 8.2 (67.4 PV ms) which value 97 out of 100 subjects exceeded.
(P300 area):
Fig. 3. The distribution pattern of square-root transformed P300 area values is displayed in the form of histogram along with the estimated curve of Gaussian distribution. The closed circle indicates the mean, with the closed star indicating mean - 2 S.D.
The P3QO-peak amplitudesand iatencies The means and SD. values of P300-peak amplitudes and latencies were 8.5 + 4.7 PV and 343.3 f 31.6 ms, respectively. The kurtosis and skewness of P300-peak amplitudes distribution were 3.821 and 0.829, respectively. While, the kurtosis and skewness for P300-peak latencies were 2.473 and 0.094, respectively. Neither amplitudes nor latencies fit the Gaussian distribution model.
The distributionof 8300 areas Due to similar reasons as in Nd, the distribution pattern of square-root transformed P300-area values was examined, which was revealed to follow the Gaussian distribution (X’(5) = 7.53, 0.25 > P > 0.10: see Fig. 3). The mean and S.D. of square-root transformed P300 areas were 44.2 (1953.9 PV ms) and 14.0 (196.6 PV ms), respectively. When the lower limit of normal P300 was defined as mean -2 SD. in square-root transformed P300 areas, it was 16.2 (261.2 PV ms) which value 98 out of 100 subjects exceeded.
Fig. 2. The distribution pattern of square-root transformed Nd area values is displayed in the form of histogram along with the estimated curve of Gaussian distribution. The closed circle indicates the mean, with the closed star indicating mean - 2 SD.
Age, gender and ERPs We checked the relationship between ages of the subjects and square-root transformed Nd or P300 areas, which revealed no significant relationships. We also calculated the correlation coefficients between ages of the subjects and amplitudes or latancies of P300-peaks. Only P3OO-peak latencies exhibited a weak but significant relationship with ages (r = 0.287, P < 0.01; see Fig. 4). As is seen in Fig. 4, one year increment in age produces 0.8 ms prolongation in P300-peak latency. We also checked possible gender-related differences in Nd and P300 areas, and amplitudes and latencies of P300-peaks. Although females demonstrated relatively larger values than males in all indices of Nd areas (females, 569.2-74.4
237 P300 LATENCY AND AGE
employed, potentially allowing some of the e?rtraP300 components such as the N200 or slow wave to contaminate the P3OOs. We thought it particularly beneficial to employ such a wide range for psychiatric populations who display greater variances in P300 latencies across trials than healthy subjects do.
CN = 100) 450
"-0.60x + 317.40 r-O.287 (P-ZOOI)
425
-__
15
q
Relationship between P300 latency and age P300 latencies displayed a weak positive corre20
25
30
3s
40
45
50
55
60
AGE (YEARS)
Fig. 4. The relationship
between P300-peak latencies and ages.
PV ms; males, 539.0 + 340.1 PV ms), P300 areas (females, 2324.1 f 1139.4 PV ms; males, 1972.9 + 1361.8 PV ms), and P300-peak amplitudes (females, 9.1 + 4.1 pV; males, 7.8 f 5.2 pV), all the differences did not reach statistical significance. Females and males displayed nearly equal values of P300-peak latencies (females, 340.2 + 30.7 ms; males, 346.5 + 32.5 ms).
DISCUSSION Rationalization of employing area as an index for F3oos
In the present study, we employed areas as an index to represent P3OOs in addition to using peak-amplitudes and peak-latencies. Previous studies, including Squires et al. (1973) and Johnson and Donchin (19781, employed area as part of their P300 indices. These authors found (1) area to be less affected by variance in P300 latencies, and (2) area correlates with behavioral data better than conventionally employed amplitudes. In addition to these reasons, we would suggest that area is especially advantageous when P3OOs are applied to psychiatric populations, in whom P300 peaks will not often be identified clearly, producing difficulties in employing peak-amplitudes and latencies as indices for P3OOs. Furthermore, in the present study a rather wide definition of the latency range for P300 was
lation with age; one year increments in age produced 0.8 ms prolongations in latencies. This result is very similar to that of many previous studies and is fairly consistent with the reports of an approx. 1 ms prolongation in the P300 latency with 1 year increments in age (Goodin et al., 1978; Gordon et al., 1986; Pfefferbaum et al., 1984). Considering many previous studies which employed the conventional oddball paradigm to elicit P3OOs, the result obtained here suggests that the P300 latency prolongation with age occurs with other tasks. Relationships among performance,
Nd and P300
Hit rates for the targets in the present study did not show any significant correlations with various indices of Nd and P300. This was thought to be due to a ceiling effect in the high hit rates observed in the present study. On the other hand, reaction times displayed significant correlations with several indices for P300, but not for Nd. Therefore, we speculate Nd is not closely associated with performance levels; while, we expect P300 to vary in relation with performance. This is almost in line with previous results. Concerning negative components of ERPs, NlOO and N200 amplitudes have been reported to correlate with reaction times, but in opposite directions from one another (NtiBtinen and Gaillard, 1974; Wilkinson and Morlock, 1966). In particular they have been associated with the arousal or wakefulness levels, which seem different from Nd which is expected to correlate more with attentional functioning. Unfortunately, relationship with Nd and performance is not well documented. P300 has been reported to vary according to hit rates and reaction times. Among them, Goodin et al. illyard et al. (1970, Pfefferbaum et al.
238
(1983) and Ruchkin et al. (1980) reported P300amplitudes correlated with hit rates. Kutas et al. (1977) and Ritter et al. (1972) and others reported that P300-latencies exhibited positive correlations with reaction times. This study adds to this core of knowledge by revealing that P300 amplitudes and areas display inverse correlation with reaction times. Overall, P300 is said to be one index of cognitive functioning. Distributions of Nd and P300 in healthy subjects While distribution-patterns
239
the present study should be validated in a larger sample. REFERENCES lackwood, D.H.R. Muir, W.J. St.Clair, D.M. and Walker, .T. (1990) Genetic trait markers in schizophrenic probands and their unaffected relatives. In M. Weller (Ed.), International perspecti~~esin schkc&mia. Biological, social and epidemiologtcal findings, John Libbey, London, pp. 129-138. Emerson. J.D. and Stoto, M.A. (1983) Transforming data. In DC. Haoglin, F. Mosteller and J.W. Tukey (Eds.), Understnnding robust and ap!orotory data am&s. John Wiley & and Sons. New York. pp. 97-128. Goodin. D.S.. Squires, KC. and Starr, A. (1978) Long latency event-related components of the auditory evoked potential in dementia. Brain, 101: 635-348. Goodin, D.S., Squires, KC. and Starr, A. (1983) Variation in early and late event-related components of the auditory evoked potentials with task difficulty. Electroencephologr. Clin. Neurophysioi.., 55: 680-686.
Gordon, E., Kraiuhin, C., Harris, A., Meares. R. and Howson, A. (1986) The differential diagnosis of dementia using P3CO latency. Biol. Psychiatry, 21: 1123-1132. Hillyard, S.A., Hink, R.F., Schwent, V.L. and Picton, P.W. (1973) Electrical signs of selective attention in the human brain. Science, 182: 177-179. Hillyard, S.A., Squires, KC,; Bauer, J. and Lindsay, P. (1971) Evoked potential correlates of auditory signal detection. Science, 172: 1357-1360. Hink, R.F.. Hillyard, S.A. and Benson, P.J. (1978) Event-related brain potentials and selective attention to acoustic and phonetic cues. Biof. Psychol., 6: I-16. Johnson, R.Jr. and Donchin, E. (1978) On how P300 amplitude varies with the utility of the eliciting stimuli. Electroencephologr. Clin. Neurophysiol., 44: 424-437.
Josiassen, R.C. Roemer, R.A. Johnson, M.M. and Shagass, C (1990) Are gender differences in schizophrenia reflected in brain event-related potentials? Schirophr. Bull., 16: 229-246.
Kutas, M., McCarthy, G. and Donchin, E. (1977) Augmenting mental chronometry: The P300 as a measure of stimulus evaluation time. Science, 197: 792-795.
Magliero, A., Bashore, T.R., Coles, M.G.H. and Don&in, E. (1984) On the dependence of P300 latency on stimulus evaluation processes. Psychophysiology, 21: 171-186. Michie, P.T., Fox, A.M., Ward, P.B., Catts, S.V. and McConaghy, N. (1990) Event-related potential indices of selective attention and cortical lateralization in schizophrenia. Psychophysiology, 27: 209-227. Nitiinen, R. and Gaillard, A.W.K. (1974) The relationship between the contingent negative variation and the reaction time under prolonged experimental conditions. BioL Psychol, 1: 227-291. Niiiitlnen, R., Gaillard, A.W.K. and Mantysalo, S. (1978) Early selective-attention effect on evoked potential reinterpreted. ktn. Psychol., 42: 313-329. Niiltiinen, R., Gaillard, A.W.K. and Varey, C.A. (1981) Attention effects on auditory EPs as a function of interstimulus interval. Biol. Psychol., 13: 173-187. Pfefferbaum, A., Ford, J.M. and Johnson, R. (1983) Manipulation of P3 latency: speed vs. accuracy instruction. Ekctroencephalogr. Clin. NeurophysioL, 5.5: 18% 19’7. Pfefferbaum, A., Ford, J.M., Wenegrat, B.G., Roth, W.T. and Kopell, B.S. (1984) Clinical application of the P3 component of event-related potentials. I. Normal aging. Electroencephalogr. Chin. Neurophysiol.., 59: 85-103.
Polich, J. (1986) Normal variation of P300 from auditory stimuli. Eiectroencephalogr. Clin. Neurophysiol., 65: 236240.
Pritchard, W.S. (1986) Cognitive event-related potential correlates of schizophrenia. PsychoL Bull., 100:43-66. Ritter, W., Simson, R. and Vaughan, H.G.Jr. (1972) Association cortex potentials and reaction time in auditory discrimination. Electroencephalogr. Clin. Neurophysiol., 33: 547-555.
Ruchkin, D.S., Sutton, S., Kietzman, M.L. and Silver, K. (1980) Slow wave and P300 in signal detection. Electroencephalogr. Clin. Neurophysiol., 50: 35-47.
Squires, K.C., Hillyard, S.A. and Lindsay, P.H. (1973) Vertex potentials evoked during auditory signal detection: Relation to decision criteria. Percept. Psychophys., 14: 265-272. Sutton, S., Braren, M., Zubin, J. and John, E.R. (1965) Evoked-potential correlates of stimulus uncertainty. Science, 1.50: 1187-1188. Wilkinson, R.T. and Morlock, H.C. (1966) Auditory evoked response and reaction time. Electroencephalogr. Clin. Neurophysiol., 23: 50-56.
233
INTPSY 00414
e
an
in a
ichi Kameyama b, Shin-Ichi Niwa ‘, Kenji Itoh d, asato Fukuda ‘, Qsamu Saitoh e, Akira Iwanami f, Nakagome g and Tsukasa Sasaki g * Research Institute of Medical Engineering, Faculty of Medicine, Uniuersity of Tokyo, Tokyo (Japan:, b Pasts and Telecommukations Tokyo Hospital, Tokyo (Japan), ’ Department of Neuropsychiatry, Fatuity of Medicine, Unitiersityof Tokyo, Tokyo (Japan), d Research Institute of Logopedics and Phoniatrics, Faculty of Medicine, University of Tokyo, Tokyo (Japan), ’ National Musashi Hospital, Kodaira (Japan), f Tokyo Metropolitan Matsuzawa Hospital, Tokyo (Japan) and s Department of Psychiatry, Faculty of Medicine, Teikyo Utkersity, Tokyo (Japan, (Accepted 28 July 1992)
Key words: Event-Related
Potential; Nd; P300, Normal value; Distribution pattern
TO obtain objective criteria for assessing the attentional and cognitive functioning of psychiatric populations, we attempted to standardize values of two components in Event-Related Potentials (ERPs), namely the attention-related negative potential (Nd) and the P300, in normal populations. The study consisted of 100 healthy volunteers (50 females, 50 males) who were given the task of making dichotic syllable discriminations requiring key-press responses. Their ages ranged between 18 and 59 years (mean f S.D., 32.3 + 11.3 years). Nd was found to be maximum in the Fz region, P300 being maximum in the Pz region. The means and standard deviations of Nd and P300 areas in their maximum regions were 554.1 + 307.8 PV ms and 2148.5 + 1248.5 /.LVms, respectively. The transformation plot for symmetry indicated the suitable power of transformation to be l/2 for both Nd and P300 distributions. After being transformed into square-root values, the distribution patterns of Nd and P300 areas were examined. When the lower limit of normal values was tentatively assigned to mean - 2 S.D. using square-root transformed data for both Nd and P300,97% of the subjects were found to display values above the lower normal limit for Nd, and 98% for the P300. Neither, Nd nor P3OOareas correlated with age, while P300 latencies displayed a w lk positive correlation with age. Females displayed relatively larger values than males for Nd and P300 areas and P300-peak amplituues. However, the differences between females and males were not statistically significant. Females and males showed nearly equal P300-peak latencies.
INTRODUCTION To obtain objective criteria for assessing attentional and cognitive functioning of psychiatric populations, we attempted to standardize values of two components in Event-Related Potentials (ERPs), namely the attention-related negative potential (Nd) and the P300, in normal populations. The Nd refers to negative components of
Correspondence to: S.-I. Niwa, Department of Neuropsychiatry, Faculty of Medicine, University of Tokyo, 7-3-l Hongo, Bunkyo-ku, Tokyo, 113 Japan.
ERPs as have been reported by Naatanen et al. (198, 1981) and Hillyard et al. (1973, 19781, which start at approx. 70-100 ms, peaking at approx. 200 ms after stimulus-onset. They are thought to emerge on occasions of differential discrimination between task-relevant and taskirrelevant stimuli, thus implicating selective attentional functioning. The P300 is a complex of positive components of ERPs that are elicited by novel or rare or task-relevant stimuli at approximately 300 ms after stimulus-onset. Sutton et al. (1965) first explained the possible cognitive meaning of P300 as the resolution of uncertainties. Later, Donchin et
234
al. (1984) hypothesized that the P300 would reflect context-updating or situation-evaluation. The P300 has been regarded as a significant complex reflecting higher cognitive functions. Though we are aware that ERPs include several components other than Nd and P300, we tried to look at these two components because they are steadily and relatively easily recorded and they bear widelyaccepted psychological meanings.
SUBJECTS AND METHODS Subjects 100 healthy adult volunteers (50 male and 50 female; ages 18-59, mean 32.3 f 11.3 years) participated in this study. 94 of the subjects were right handed, 5 were mixed handed, and the remaining one subject was left handed. All of the subjects were free from any hearing disability, and had no history of neurological or psychiatric disorders. The mean of their number of years of education was 14.4 f 2.3 (range 8-22) years. The subjects consisted of 30 university students, 16 engineers, 16 laborers, 10 office workers, 6 doctors, 7 nursing staffs and 15 housewives. There were no differences between female and male subjects in their ages and years of education. Methods In the ~ic~otic syllable discrimination tasks empiuyed m the present study, male and female voices presented the syllables (/te/ and /ga/> to the subjects via heplphones, with the male voice being presented to one ear and the female voice to the other ear with no changes between the stimuli within a session. The frequency-ratio in appearance of /te/ and /ga/ was 1:3. The stimuli were applied in a random, alternative pattern to one side or the other and to only one ear at a time during one session. Subjects were instructed to press a designated response-key to the target with the index or middle finger of the attended side hand, while pressing another key to the other stimuli with the other side hand. In the instruction, both accuracy and speed of responses were equally required. The duration of each stimulus was 150 ms; the stimulus intensity being set
approx. at 60 dBSL. The inter-stimulus intervals -2000 ms. AII of varied randomly between 1 the subjects performed 4 runs, that is, one target syllable X two voices (male or female) X two attended channels (left or right ear). The number of the total stimuli in both ears for each run was approx. 240. The subjects were seated in a sound-proof anechoic room with eyes closed. According to the International lo-20 Electrode System, EEGs were derived from the Fz, Cz and Pz regions monopolarly referenced to linked ear-lobe electrodes. Eye-movements were also recorded bipolarly between above and lateral electrodes to the right eye. EEGs were amplified with bandpass down 6 dB at 0.15 and 300 Hz. After processed through an anti-aliasing filter of approximately 95 Hz, EEGs were digitized on-line by micro PDP 1 l/23 with a sampling frequency of 250 Hz/Ch. EEG data with amplitudes exceeding 100 PV measured between the most negative peak and the most positive through or accompanied by EOGs of more than 150 PV during the period 40 ms prestimulus to 800 ms post-stimulus were eliminated from averaging. In the population employed in the present study, the actual number of rejected EEG responses from averaging remained small. After a smoothing process using a digital filter with a window-width of 32 ms (8 data points), the averaged ERP waveforms for the individual subjects were averaged separately through VAX 8530 into four catories: (1) target syllables in the attended ear; (2) non-target syllables in the attended ear; (3) target syllables in the non-attended ear; and (4) non-target syllables in the nonattended ear. The averaging times for /te/ and /ga/ in one ear were limited to 16 and 48, respectively, for each run, hence producing 64 times for categories 1 and 3 and 192 times for categories 2 and 4 in total, being multiplying by 2 voices X 2 sides. Specifically, first 16 responses for categories 1 and 3 and first 48 responses for 2 and 4 were averaged for each run. It is empirically reasonable to assume that psychiatric patients tend to produce fewer numbers of employable EEG responses for averaging as compared to healthy subjects. The reason of the restriction in averaging times mentioned above is this empir-
ical assumption that requires a balanced averaging times between healthy subjects and pschiatric populations. The Nd was defined as the negative component of a difference wave, in the period of O-400 ms after the stimulus onset, between ERPs ited by non-target syllables in the attended ear and those for non-target syllables in the nonattended ear. For the index to represent Nd, those areas were employed which were defined as the negative areas between O-400 ms in the difference wave described above. P300 was defined as the positive component of ERP, in the period of 260-m ms after the stimulus onset. P300 areas, P300 amplitudes and P300 latencies for targets in the attended ear were employed as indices representing P300. P300 areas were defined as the positive areas of target ERPs between 260-600 ms, with P300 amplitudes and P300 latencies corresponding to those measures of the most positive peaks in the relevant latency periods. Amplitudes were measured with reference to the 0 level defined as the mean amplitude during the 49 ms period just prior to the stimulus onset. RESULTS ERP wacefmms
The grand average of ERP waveforms for one hundred subjects are shown in Fig. 1. Waveforms for the four stimulus categories are shown separately in the figure along with the wave differences. The P3OOsin Fig. 1 are most clearly elicited by the targets in the attended ear particularly at the Pz region. Fig. 1 also reveals that Nd waves are most prominent at the Fz region in which two peaks are identified. In the Cs and Pz regions, the first peak, Cz, is clearer than the second one. The Nd for the Fz region and P300 for the Pz region are utilized in the following analyses. Perforrmxc
and ERPs
Reaction tire-:s and hit rates for the targets were measured in 91 subjects (46 males, 45 females). Indices for the remaining 9 subjects were not measured due to failures in the experimental apparatus. The mean and S.D. of ages for the 91
-1OjfV
k
0
*
-
- 4c(V 0
* t-Q--
-_.__;:-
-
23.5
topv-
?
_.__..- Oi
___---
:.*-r“.., . ‘.. ‘C’_--..*
*
J’-..‘._
‘._.--
*-
__
_A-.
Fz ......... cz --- h
_-
I I I I I I
0
sxhl!T
Fig. 1. The waveforms for the four stimulus categories are separately shown along with the difference waves. Upper left, attend non-target; upper right, attend target; middle left. non-attend non-target; lower right, non-attend target; lower left, difference wave. FL Cz and Pz denote the electrode locations.
subjects was 32.8 f 11.5 years, with these values almost similar to those of the whole subjects. Hit rates for the targets ranged between 85.8 to 100% (mean f S-D., 97.4 f 2.9), with the mean reaction times for individual subjects distributing between 322 to 844 ms (mean + S.D., 545 f 87 ms). Possible correlations between hit rates (or reaction times) and idd (or P300) were examined through calculation of partial correlation coefficients controlling for ages, which revealed the following results: (1) hit rates showed no correlation with the Nd area, P300 area, P300 amplitude and P300 latency. (2) reaction times displayed no correlation with the Nd area, while they demonstrated weak !-.:_:a significant correlations with the P300 area (r = -0.37, P < O.OOl), P300 latency (r = 0.24, P < 0.05) and P300 amplitude (r = - 0.42, P < 0.001). Relationship between the Nd and P300
The means and S.D. of the Nd area at Fz and P300 at Pz were 554.1 f 307.8 PV ms and 2148.5 + 1261.5 PV ms, respectively. The relationship between Nd and P300 areas was not significant (Pearson’s correlation coefficient r = 0.07, n.s.1. The distribution of Nd areas
Since the distributira pattern of absolute Nd area values was asymmetric, with the number of
236
smaller Nd subjects exceeding that of larger Nd subjects (skewness = 0.296), the suitable power of transformation was obtained according to the method of the transformation plot for symmetry (Emerson and Stoto, 1983). This method indicated the power to be l/2. The distribution pattern of square-root transformed Nd values was obtained, of which histogram is displayed in Fig. owever, this distribution reached statistical significance not to follow the Gaussian distribution (X2(5) = 14.24, 0.01 < P < 0.025). The mean and SD. of square-root transformed Nd areas were 22.4 (502.7 PV ms) and 7.1 (50.4 PV ms), respectively. When the lower limit of normal Nd was tentatively defined as mean -2 SD. in square-root transformed Nd areas, it was 8.2 (67.4 PV ms) which value 97 out of 100 subjects exceeded.
(P300 area):
Fig. 3. The distribution pattern of square-root transformed P300 area values is displayed in the form of histogram along with the estimated curve of Gaussian distribution. The closed circle indicates the mean, with the closed star indicating mean - 2 S.D.
The P3QO-peak amplitudesand iatencies The means and SD. values of P300-peak amplitudes and latencies were 8.5 + 4.7 PV and 343.3 f 31.6 ms, respectively. The kurtosis and skewness of P300-peak amplitudes distribution were 3.821 and 0.829, respectively. While, the kurtosis and skewness for P300-peak latencies were 2.473 and 0.094, respectively. Neither amplitudes nor latencies fit the Gaussian distribution model.
The distributionof 8300 areas Due to similar reasons as in Nd, the distribution pattern of square-root transformed P300-area values was examined, which was revealed to follow the Gaussian distribution (X’(5) = 7.53, 0.25 > P > 0.10: see Fig. 3). The mean and S.D. of square-root transformed P300 areas were 44.2 (1953.9 PV ms) and 14.0 (196.6 PV ms), respectively. When the lower limit of normal P300 was defined as mean -2 SD. in square-root transformed P300 areas, it was 16.2 (261.2 PV ms) which value 98 out of 100 subjects exceeded.
Fig. 2. The distribution pattern of square-root transformed Nd area values is displayed in the form of histogram along with the estimated curve of Gaussian distribution. The closed circle indicates the mean, with the closed star indicating mean - 2 SD.
Age, gender and ERPs We checked the relationship between ages of the subjects and square-root transformed Nd or P300 areas, which revealed no significant relationships. We also calculated the correlation coefficients between ages of the subjects and amplitudes or latancies of P300-peaks. Only P3OO-peak latencies exhibited a weak but significant relationship with ages (r = 0.287, P < 0.01; see Fig. 4). As is seen in Fig. 4, one year increment in age produces 0.8 ms prolongation in P300-peak latency. We also checked possible gender-related differences in Nd and P300 areas, and amplitudes and latencies of P300-peaks. Although females demonstrated relatively larger values than males in all indices of Nd areas (females, 569.2-74.4
237 P300 LATENCY AND AGE
employed, potentially allowing some of the e?rtraP300 components such as the N200 or slow wave to contaminate the P3OOs. We thought it particularly beneficial to employ such a wide range for psychiatric populations who display greater variances in P300 latencies across trials than healthy subjects do.
CN = 100) 450
"-0.60x + 317.40 r-O.287 (P-ZOOI)
425
-__
15
q
Relationship between P300 latency and age P300 latencies displayed a weak positive corre20
25
30
3s
40
45
50
55
60
AGE (YEARS)
Fig. 4. The relationship
between P300-peak latencies and ages.
PV ms; males, 539.0 + 340.1 PV ms), P300 areas (females, 2324.1 f 1139.4 PV ms; males, 1972.9 + 1361.8 PV ms), and P300-peak amplitudes (females, 9.1 + 4.1 pV; males, 7.8 f 5.2 pV), all the differences did not reach statistical significance. Females and males displayed nearly equal values of P300-peak latencies (females, 340.2 + 30.7 ms; males, 346.5 + 32.5 ms).
DISCUSSION Rationalization of employing area as an index for F3oos
In the present study, we employed areas as an index to represent P3OOs in addition to using peak-amplitudes and peak-latencies. Previous studies, including Squires et al. (1973) and Johnson and Donchin (19781, employed area as part of their P300 indices. These authors found (1) area to be less affected by variance in P300 latencies, and (2) area correlates with behavioral data better than conventionally employed amplitudes. In addition to these reasons, we would suggest that area is especially advantageous when P3OOs are applied to psychiatric populations, in whom P300 peaks will not often be identified clearly, producing difficulties in employing peak-amplitudes and latencies as indices for P3OOs. Furthermore, in the present study a rather wide definition of the latency range for P300 was
lation with age; one year increments in age produced 0.8 ms prolongations in latencies. This result is very similar to that of many previous studies and is fairly consistent with the reports of an approx. 1 ms prolongation in the P300 latency with 1 year increments in age (Goodin et al., 1978; Gordon et al., 1986; Pfefferbaum et al., 1984). Considering many previous studies which employed the conventional oddball paradigm to elicit P3OOs, the result obtained here suggests that the P300 latency prolongation with age occurs with other tasks. Relationships among performance,
Nd and P300
Hit rates for the targets in the present study did not show any significant correlations with various indices of Nd and P300. This was thought to be due to a ceiling effect in the high hit rates observed in the present study. On the other hand, reaction times displayed significant correlations with several indices for P300, but not for Nd. Therefore, we speculate Nd is not closely associated with performance levels; while, we expect P300 to vary in relation with performance. This is almost in line with previous results. Concerning negative components of ERPs, NlOO and N200 amplitudes have been reported to correlate with reaction times, but in opposite directions from one another (NtiBtinen and Gaillard, 1974; Wilkinson and Morlock, 1966). In particular they have been associated with the arousal or wakefulness levels, which seem different from Nd which is expected to correlate more with attentional functioning. Unfortunately, relationship with Nd and performance is not well documented. P300 has been reported to vary according to hit rates and reaction times. Among them, Goodin et al. illyard et al. (1970, Pfefferbaum et al.
238
(1983) and Ruchkin et al. (1980) reported P300amplitudes correlated with hit rates. Kutas et al. (1977) and Ritter et al. (1972) and others reported that P300-latencies exhibited positive correlations with reaction times. This study adds to this core of knowledge by revealing that P300 amplitudes and areas display inverse correlation with reaction times. Overall, P300 is said to be one index of cognitive functioning. Distributions of Nd and P300 in healthy subjects While distribution-patterns
239
the present study should be validated in a larger sample. REFERENCES lackwood, D.H.R. Muir, W.J. St.Clair, D.M. and Walker, .T. (1990) Genetic trait markers in schizophrenic probands and their unaffected relatives. In M. Weller (Ed.), International perspecti~~esin schkc&mia. Biological, social and epidemiologtcal findings, John Libbey, London, pp. 129-138. Emerson. J.D. and Stoto, M.A. (1983) Transforming data. In DC. Haoglin, F. Mosteller and J.W. Tukey (Eds.), Understnnding robust and ap!orotory data am&s. John Wiley & and Sons. New York. pp. 97-128. Goodin. D.S.. Squires, KC. and Starr, A. (1978) Long latency event-related components of the auditory evoked potential in dementia. Brain, 101: 635-348. Goodin, D.S., Squires, KC. and Starr, A. (1983) Variation in early and late event-related components of the auditory evoked potentials with task difficulty. Electroencephologr. Clin. Neurophysioi.., 55: 680-686.
Gordon, E., Kraiuhin, C., Harris, A., Meares. R. and Howson, A. (1986) The differential diagnosis of dementia using P3CO latency. Biol. Psychiatry, 21: 1123-1132. Hillyard, S.A., Hink, R.F., Schwent, V.L. and Picton, P.W. (1973) Electrical signs of selective attention in the human brain. Science, 182: 177-179. Hillyard, S.A., Squires, KC,; Bauer, J. and Lindsay, P. (1971) Evoked potential correlates of auditory signal detection. Science, 172: 1357-1360. Hink, R.F.. Hillyard, S.A. and Benson, P.J. (1978) Event-related brain potentials and selective attention to acoustic and phonetic cues. Biof. Psychol., 6: I-16. Johnson, R.Jr. and Donchin, E. (1978) On how P300 amplitude varies with the utility of the eliciting stimuli. Electroencephologr. Clin. Neurophysiol., 44: 424-437.
Josiassen, R.C. Roemer, R.A. Johnson, M.M. and Shagass, C (1990) Are gender differences in schizophrenia reflected in brain event-related potentials? Schirophr. Bull., 16: 229-246.
Kutas, M., McCarthy, G. and Donchin, E. (1977) Augmenting mental chronometry: The P300 as a measure of stimulus evaluation time. Science, 197: 792-795.
Magliero, A., Bashore, T.R., Coles, M.G.H. and Don&in, E. (1984) On the dependence of P300 latency on stimulus evaluation processes. Psychophysiology, 21: 171-186. Michie, P.T., Fox, A.M., Ward, P.B., Catts, S.V. and McConaghy, N. (1990) Event-related potential indices of selective attention and cortical lateralization in schizophrenia. Psychophysiology, 27: 209-227. Nitiinen, R. and Gaillard, A.W.K. (1974) The relationship between the contingent negative variation and the reaction time under prolonged experimental conditions. BioL Psychol, 1: 227-291. Niiiitlnen, R., Gaillard, A.W.K. and Mantysalo, S. (1978) Early selective-attention effect on evoked potential reinterpreted. ktn. Psychol., 42: 313-329. Niiltiinen, R., Gaillard, A.W.K. and Varey, C.A. (1981) Attention effects on auditory EPs as a function of interstimulus interval. Biol. Psychol., 13: 173-187. Pfefferbaum, A., Ford, J.M. and Johnson, R. (1983) Manipulation of P3 latency: speed vs. accuracy instruction. Ekctroencephalogr. Clin. NeurophysioL, 5.5: 18% 19’7. Pfefferbaum, A., Ford, J.M., Wenegrat, B.G., Roth, W.T. and Kopell, B.S. (1984) Clinical application of the P3 component of event-related potentials. I. Normal aging. Electroencephalogr. Chin. Neurophysiol.., 59: 85-103.
Polich, J. (1986) Normal variation of P300 from auditory stimuli. Eiectroencephalogr. Clin. Neurophysiol., 65: 236240.
Pritchard, W.S. (1986) Cognitive event-related potential correlates of schizophrenia. PsychoL Bull., 100:43-66. Ritter, W., Simson, R. and Vaughan, H.G.Jr. (1972) Association cortex potentials and reaction time in auditory discrimination. Electroencephalogr. Clin. Neurophysiol., 33: 547-555.
Ruchkin, D.S., Sutton, S., Kietzman, M.L. and Silver, K. (1980) Slow wave and P300 in signal detection. Electroencephalogr. Clin. Neurophysiol., 50: 35-47.
Squires, K.C., Hillyard, S.A. and Lindsay, P.H. (1973) Vertex potentials evoked during auditory signal detection: Relation to decision criteria. Percept. Psychophys., 14: 265-272. Sutton, S., Braren, M., Zubin, J. and John, E.R. (1965) Evoked-potential correlates of stimulus uncertainty. Science, 1.50: 1187-1188. Wilkinson, R.T. and Morlock, H.C. (1966) Auditory evoked response and reaction time. Electroencephalogr. Clin. Neurophysiol., 23: 50-56.
