Sei sulla pagina 1di 9

Psychophysiology, 36 ~1999!, 543 – 551. Cambridge University Press. Printed in the USA.

Copyright © 1999 Society for Psychophysiological Research

Dynamics of SCR, EEG, and ERP activity in an oddball

paradigm with short interstimulus intervals


Department of Neurology, University of Sydney and Westmead Hospital, Sydney, Australia
Cognitive Neuroscience Unit, Department of Psychological Medicine, University of Sydney and Westmead Hospital,
Sydney, Australia
Department of Medical Physics, University of Sydney and Westmead Hospital, Sydney, Australia
Brain Dynamics Laboratory, Mental Health Research Institute of Victoria, Australia

Studies of concurrent central, and autonomic activity using a conventional event-related potential ~ERP! oddball
paradigms, are considered useful in elucidating the relationship between central and autonomic responses, but the
autonomic response tends to overlap. A new method was used to decompose and score overlapping skin conductance
responses ~SCR!. This method enabled examination of dynamic relationships of phasic SCR, prestimulus electro-
encephalogram ~EEG!, and ERP to auditory target stimuli in 50 normal adults. SCR amplitude was negatively correlated
to EEG and N200 amplitude. The SCR amplitude changes over time exhibited an exponential decline opposite to those
of N200, alpha, and beta. All the fitted exponential functions had a time constant of 1–2 min. The findings suggest that
a N200 component, active in the auditory sensory discrimination, is concomitant with the SCR. The narrow range of
the time constant may provide a clue to the conjoint processes underlying central and autonomic adaptive functions.
Descriptors: EEG, N2b, SCR, Electrodermal activity, Orienting reflex, Time constant

Research studies on electrodermal activity ~EDA!, electroenceph- Vasilieva, 1975!. A functional link between EEG and electroder-
alography ~EEG!, and event-related evoked potential ~ERP!, have mal activity, as indices of arousal and orienting is expected, but
become essentially separate disciplines since their respective in- poststimulus EEG contains both ERD and ERP. Prestimulus EEG
ception in 1888 ~Bloch, 1993!, 1929 ~Brazier, 1961; Gloor, 1969!, measures, on the other hand, can be used as an index of the brain
and 1964 ~Halliday, 1980!. This paper serves to: ~a! briefly re- state modulating ERP responses ~Basar, Basar-Eroglu, Rosen, &
view possible common adaptive functions of EEG, ERP, and SCR; Schütt, 1984; Intriligator & Polich, 1994!. Prestimulus alpha and
~b! present the difficulties of concurrent central nervous system beta activity have been associated with prestimulus electrodermal
~CNS! and autonomic nervous system ~ANS! studies; ~c! describe level modulated by arousal ~Lim et al., 1996!.
a solution to these difficulties; and, ~d! report for the first time the
interrelationships between concomitant CNS–ANS variables in a ERP
paradigm with short interstimulus intervals ~ISI!. Multiple components have been found to be active around the time
of N100 ~Näätänen & Picton, 1987!. N100 is generally regarded as
EEG an exogenous component of ERP modulated by stimulus param-
The possible genesis of various rhythmic EEG activities have been eters and attention. Enhancement of ERP late components ~N200
reviewed extensively ~Lopes da Silva, 1991; Steriade, Gloor, Llinàs, and P300! to stimuli with temporal uncertainty ~Sutton, Baren,
Lopes da Silva, & Mesulam, 1990!. Variation of brainstem tonic Zubin, & John, 1965!, with an unpredicted sudden change ~Ritter,
excitation to the cortical neurons leads to arousal, EEG desynchro- Vaughan, & Costa, 1968!, and to selective attention ~Ford, Roth,
nization, and shifts in EEG frequency ~Cobb, 1963; Kellaway, Dirks, & Kopell, 1973; Hillyard, Hink, Schwent, & Picton, 1973!
1990; Montplaisir, 1975; Moruzzi & Magoun, 1949; Niedermeyer, have been demonstrated. N200 has been suggested to reflect a
1987!. Event-related desynchronization ~ERD! following auditory central orienting response ~Ford, Roth, & Kopell, 1976; Squires,
or visual stimulation has also been demonstrated ~Pfurtscheller & Squires, & Hillyard, 1975! and is thought to reflect a decision
Aranibar, 1977!, and poststimulus alpha blockage ~desynchroniza- process related to sensory discrimination of attended stimuli ~Rit-
tion! has been linked to the orienting reflex ~Voronin, Bonfitto, & ter, Simon, Vaughan, & Friedman, 1979!. In the time zone of
N200, there are two recognized components, N2a and N2b ~Per-
rault & Picton, 1984!. N2a is regarded as an automatic response to
Address reprint requests to: Chong Lee Lim, Neurology Department,
deviant stimuli in a series of standard stimuli with its maximum
Westmead Hospital, Westmead 2145, Australia. E-mail: lee@neuro.wsahs. amplitude over the anterior head region ~Näätänen & Gaillard, 1983!. A subset of N2a, the mismatch negativity ~MMN!, has been

544 C.L. Lim et al.

proposed as the cerebral initiation of the orienting reflex ~OR; which SCR is a principal component ~Boucsein, 1992, p. 218!. The
Näätänen, 1992!. SCR habituation associated with repetitive, regular, unattended
N2b is a broad, centrally distributed wave between 150 and stimuli is well known ~Boucsein, 1992; Graham, 1973; Roy, Se-
215 ms in response to infrequent attended auditory stimuli. N2b queira, & Delerm, 1993; Siddle, Remington, Kuiack, & Haines,
has been interpreted as a brain event associated with transient 1983; Siddle, Smith, & Marcer, 1974!. Most electrodermal studies
arousal and the OR ~Loveless, 1983; Näätänen & Gaillard, 1983!. have used paradigms of long ISIs. Recently, SCR habituation in
In attend oddball conditions, N2b is often followed by P3a ~fron- the nonattend condition has been shown using a very short ISI of
tocentrally distributed, peaking between 220 and 280 ms!, also 1.1 s ~Barry, Feldmann, Gordon, Cocker, & Rennie, 1993!. The
associated with the OR ~Squires et al., 1975!. The N2b–P3a wave demonstration suggests that the adaptive character of SCR is not
complex has been widely linked to the orienting response ~Halgren compromised with the use of short ISI stimulation.
& Marinkovic, 1994; Loveless, 1983; Lyytinen & Näätänen, 1987;
Näätänen & Gaillard, 1983!. CNS–ANS Integration
P3b ~P3 or P300! has a longer latency and is more prominent Concurrent measures of central and autonomic function have proved
in the parietal region. It is thought to reflect “context updating” difficult because ERPs have a time course in the order of a fraction
~Donchin, 1979!, or “context closure” ~Verleger, 1988!. P3b has of a second, whereas the SCR has a time course in the order of
not been shown to habituate and it is, therefore, not considered a several seconds ~and SCL varies over an even longer time scale!.
likely correlate of autonomic OR ~Donchin, 1981; Donchin et al., A paradigm with a short ISI, such as conventionally used in ERP
1984!. These findings indicate that any central adaptive changes research, results in overlapping SCRs ~Figure 1!, making quanti-
that relate to electrodermal activity are most likely to be seen in the tation of responses difficult ~Boucsein, 1992; Dawson, Schell, &
time zone of N200. Filion, 1990!. This difficulty has contributed to the paucity of
research into concurrent ANS and CNS activities.
SCR Some ANS and CNS studies have employed intermediate ISIs
Skin conductance ~SC! reflects the activity of the eccrine sweat ~Roth, Goodale, & Pfefferbaum, 1991!, whereas others used ANS
glands innervated by the sympathetic sudomotor nerves of the data qualitatively to study EEG ~Davies & Krkovic, 1964; Voronin
autonomic nervous system ~Fowles, 1986; Wallin, 1981!. SC is et al., 1975! or ERP activity ~Lyytinen, Blomberg, & Näätänen,
composed of two components—a phasic, stimulus-related SCR, 1992!. Short ISI auditory stimulation is known to be more useful
and a tonic arousal-related skin conductance level ~SCL! ~Bouc- for study of focused attention ~Schwent, Hillyard, & Galambos,
sein, 1992!. SCR waveform is considered to reflect sweat activity 1976!, and alpha and galvanic skin responses ~GSR! to light stim-
in the sweat ducts in the skin and the opening and the collapse of uli of ISI of 2–3 s show closer correspondence than those respond-
the sweat duct near the pores ~Edelberg, 1993!. SCR is regarded as ing to stimuli of longer ISI ~Voronin et al., 1975!. Therefore, a
the most useful laboratory test for an autonomic response ~Dam- conventional paradigm with short ISI was used in this study.
asio, 1994!. SCR is a widely investigated autonomic response Subaveraged ERPs have long been used in ERP studies to
~Graham, 1973! and is usually regarded as an index of OR ~Fowles, uncover ERP features that would otherwise be obscured by the
1986; Näätänen, 1992, p. 63!. Sokolov ~1960! regarded what Pav- conventional averaging ~Ritter, Simon, Vaughan, & Friedman, 1979;
lov ~1927! described as a generalized aroused state following an Starr, Sandroni, & Michalewski, 1995!. Few studies have assessed
unexpected external stimulus as a manifestation of a generalized the dynamic of central and autonomic changes across a trial. This
OR, and later extended it to include focal OR ~Sokolov, 1975!, of was the focus of the current study. We predicted that:

Figure 1. Skin conductance traces ~upper! from an individual attending and responding to 40 targets presented at random but fixed
sequential times ~upward markers on the time axis! during a trial. The responses to the targets were complex, consisting of numerous
overlapping skin conductance responses, each with a fast rise and a slow decline.
Dynamics of concurrent CNS and ANS activity 545

1. SCR responsivity may increase ~Roth et al., 1991! and SCR The electrodes were excited by a constant voltage of 0.5 V ~Fowles
amplitude may decrease with repetition of the target stimuli et al., 1981; Lykken & Venables, 1971! so that the recorded signal,
across the trial ~Barry et al., 1993!. proportional to the resultant current, was also proportional to the
2. Amplitude of prestimulus EEG alpha activity and beta activity
may be low at the beginning of the trial and enhanced toward Skin Conductance Response Evaluation System (SCORES)
the end ~Lim et al., 1996!. SCORES scans the entire SC data of a session and for each re-
3. ERP peak components in the latent times of N200, and perhaps sponse complex associated with a stimulus, iteratively optimizes
early P300, may be modified by stimulus repetition ~Näätänen, the model parameters so as to minimize the mean square difference
1982; Näätänen & Gaillard, 1983; Picton, Hillyard, Krauz, & between the actual and the fitted curves. The fitted curves have
Galambos, 1974; Squires et al., 1975!. been shown to be almost indistinguishable from the actual raw
response complex ~the top two superimposed SCR curves in Fig-
4. The concurrent central and autonomic measures are likely to ure 2!. At this point ~for an example see Figure 2!, in the vast
show some systematic change over the trial. majority of cases, the mathematically generated pattern fitted the
actual raw response complex well. None had residuals exceeding
Method 5% of the response amplitude. The decomposed parts are the tonic
level, the decaying residual from the last response, and the two
Subjects “pure” SCRs ~SCR 1 and SCR 2! displaced by time interval of two
Fifty normal subjects ~25 men and 25 women aged 18–70 years, stimuli ~the four parts below the superimposed raw and fitted
mean 5 45.4 years, SD 5 15.0 years! participated in the study. curves in Figure 2!. A range of variables including peak amplitude
Subjects were recruited from the surrounding community. Subjects and latency of the first “pure” response associated with a stimulus
with a history of a major medical, neurological, or psychiatric ~SCR 1! are measured automatically. More details regarding the
disorder or with history of alcohol or other drug abuse were ex- mathematics are available elsewhere ~appendix in Lim et al., 1997!.
cluded. Subjects were asked to avoid tobacco or caffeinated drinks
for 3 hr and alcoholic beverages for 1 day before the recording

Stimulus Paradigm and Data Acquisition

The subjects were seated comfortably in a quiet, dimly lit labora-
tory with air-conditioned ambient temperature set at 24 6 18C.
Subjects sat facing a monitor screen and wearing a pair of head-
phones. The operator was in an adjacent control room with audio-
visual links. A conventional auditory oddball paradigm was used,
with a constant ISI of 1.32 s. The hearing threshold was deter-
mined. Auditory stimulation consisted of tones at 60 dB above
threshold with duration 50 ms and with a rise and a fall time of
10 ms. Tones of 1000 Hz were designated as background and tones
of 1500 Hz as target. A total of 40 target and 247 background tones
were presented in a random but fixed sequence, with intertarget
intervals ~ITI! varying between 2.64 and 15.84 s. Subjects were
instructed to attend to the target tone and respond with a button
press “as quickly as possible” when they were “sure” they had
heard a target stimulus. They were also asked to look at a colored
dot on the center of the video screen. SC, EEG, single trial ERP,
and electrooculogram ~EOG! were recorded simultaneously using
a 32-channel Syn Amps DC amplifier system ~Neuro Scan Inc.,
Sterling, Virginia, USA!. The horizontal EOG was recorded via a
pair of electrodes placed 1 cm lateral to the outer canthi and the
vertical EOG was monitored via a pair of electrodes placed above
and below the right eye in line with the center of the pupil. The
EOG signals were used for correction of their contribution to the
scalp recorded EEG by a technique based on Gratton, Coles, and
Donchin ~1983!. The signals were band-limited to 50 Hz, digitized
at 250 Hz, and stored continuously for more than 6.5 min for sub-
sequent offline processing.
Figure 2. An amplified version of two typical overlapping skin conduc-
tance responses ~SCRs! from an individual, which has been processed by
SC Recording
our software package, SCORES. The raw and the fitted skin conductance
SC was recorded by a pair of silver-silver chloride electrodes, curves are superimposed and almost indistinguishable from each other
approximately 0.8 cm 2 in contact area, filled with electrode paste ~top!. The flat line below them is the skin conductance level. The three
of 0.05 M sodium chloride in an inert ointment and placed on the decomposed traces ~bottom! from left to right are the residuals from a
volar surface of the distal phalanges of digits II and III of the previous response, the SCR in response to a target ~presented at 0 second!
nondominant hand. The skin area was wiped with an alcohol swab. and another SCR responding to the next target ~presented at 2.64 s!.
546 C.L. Lim et al.

All 10-s epochs containing an SCR within 1–3 s following each the trial. Temporal trends of these measures were curve-fitted using
of the 40 targets in each subject were evaluated. The segments a standard nonlinear least-squares routine known as the Marquardt-
containing composite SC signals and often overlapping responses Levenberg method ~Press, Flannery, Teukolsky, & Vetterling, 1992!.
were decomposed into phasic SCR and tonic SCL components by Before statistical analysis, SCR amplitudes were transformed log-
curve-fitting using our software package SCORES. The SCL and arithmically to reduce skewness in the distribution and to improve
the “pure” SCR amplitude and latency associated with each target normality. To avoid undue influence of individual spurious mea-
stimulus were obtained, sorted by target and averaged across sub- surements and to satisfy the normality assumption of the analysis
jects. The four parameter values of all the SCRs were grouped into of variance ~ANOVA!, outliers were removed. For each of the
40 target bins and subaveraged ~missing values did not contribute three variables, cases in each target with values greater ~or less!
to the average!. An SCR waveform was constructed from the four than 1.5 times the interquartile range from the upper ~or lower!
parameter values for each target for the group. quartile were considered outliers ~0.82% of alpha data, 0.92% of
beta data, 1.08% of SCR data!. Separate correlation studies of SCR
EEG amplitudes versus total alpha, beta, and N200 amplitude data over
EEG was recorded from 19 head sites placed according to the trial were performed.
10-20 system ~Jasper, 1958! in reference to linked-ear electrodes.
The EEG was amplified by 200, digitized, and stored continuously
during the session. After EOG correction, any prestimulus EEG
epoch containing signal in excess of 6500 mV was excluded from The fact that the decomposed parts of the SCR summated to the
analysis. Fast Fourier transforms ~FFT! of 1-s EEG segments prior raw trace with little error suggests that the overlapping SCRs ~Fig-
to each target were then performed after applying a windowing ure 1! had been scored successfully ~Figure 2!. A total of 1,421
function to minimize leakage and end effects. The EEG spectral SCRs ~or 71% of targets! were obtained from all 50 subjects. SCR,
amplitudes from 0 to 48 Hz were grouped into sub-bands as fol- N200, and EEG frequency spectra showed systematic dynamic
lows: delta band ~1–3 Hz!, theta band ~4–7 Hz!, alpha band 1 patterns from the 1st to the 40th target stimulus ~Figure 3!. The
~8–9 Hz!, alpha band 2 ~10–13 Hz!, alpha ~8–13 Hz!, beta band 1 first SCR was by far the largest, followed by the next six target
~14–18 Hz!, beta band 2 ~19–24 Hz!, beta band 3 ~25–30 Hz!, and responses, which were larger than the rest ~Figure 3A!. Alpha band
beta band 4 ~31– 48 Hz!. EEG data from only Cz site was analyzed. amplitude showed the most prominent changes among the four
As the delta band, theta band, and beta band 4 data did not show EEG bands. Alpha amplitude was initially low ~first three targets!
a systematic changes over target presentation times, no further and became increasingly larger up to the 25th target and varied in
analysis was performed on them. The two alpha bands 1 and 2 between these levels thereafter ~Figure 3B!. Beta changes fol-
exhibiting similar trends were lumped together and averaged as a lowed a similar pattern, but were less obvious because of their
single alpha band. Similarly, beta bands 1, 2, and 3 were grouped low amplitude relative to the alpha band. The N200 amplitude was
into a single beta band. The root mean squares ~rms! amplitudes the only ERP component that showed a systematic change over the
were used for trend analysis. The alpha and beta rms amplitudes trial ~Figure 3C!. N200 amplitude increased markedly after the
associated with each target stimulus were averaged across subjects. first five targets.
Three key features of the CNS and ANS activity were identi-
ERP fied. ~1! The dynamic amplitude changes of SCR, alpha, beta, and
Target ERPs at Cz were extracted from the continuous recordings, N200 were best modeled as exponential functions of the time of
with each epoch extending from 200 ms prestimulus to 800 ms target presentation. The parameters of the exponential time func-
poststimulus. The average of the 200-ms prestimulus EEG served tions are listed in Table 1. SCR amplitude exhibited an expo-
as the baseline for amplitude measurement. Target ERPs were nential decline with time with a high goodness of fit ~Figure 4A,
subjected to the same artifact rejection method as employed for Table 1!. There was an exponential rise from an initial low am-
EEGs. The single trial epochs associated with each of the 40 tar- plitude, both in EEG alpha and beta amplitudes over the trial
gets were averaged across subjects, resulting in an ensemble of 40 ~Figure 4B! and N200 ERP ~Figure 4C, Table 1!. ~2! SCR, alpha,
ERPs corresponding to targets 1 to 40. N100, P200, N200, and beta, and N200 dynamic profile over the trial reached a plateau
P300 ERP peak latencies and amplitudes falling within the follow- within 3 min and had time constants between 1 and 2 min ~Fig-
ing windows ~80–140 ms for N100; 150–200 ms for P200; 180– ure 3, Table 1!. ~3! Log ~SCR! was significantly correlated indi-
280 ms for N200; 280–550 ms for P300! were scored automatically. vidually with alpha, beta and N200 amplitude ~Table 2!.
The peak within each window was determined by the signs of the
gradients before and after a specific data point. The specific data
point was defined as the peak if one sign of its preceding gradients Table 1. The Variables and Parameters of the Fitted
sustained for more than three data points, and an opposite sign of Exponential Time Functions
its ensuing gradients also sustained for more than three data points.
The peak amplitude was measured from a baseline defined as the Time Initial
mean of the data over a 200-ms prestimulus period. The peak Dependent constant ~TC! level Constant goodness-of-fit
latency was determined from the stimulus onset to the peak time. variable ~s! ~a0 ! ~c! ~r 2 !
The automatic scores were checked by two reviewers and were SCR ~mS! 59.6 0.25 0.21 0.82
re-scored when required in a small number of cases. EEG alpha ~mV! 70.1 1.06 2.26 0.59
EEG beta ~mV! 101.6 0.82 2.76 0.54
Statistics and Analysis ERP N200 ~mV! 55.8 7.86 0.00 0.49
Preliminary exploratory analysis of the trends of all measured
variables for each targets revealed that only SCR, alpha amplitude, Note: SCR 5 skin conductance response; EEG 5 electroencephalogram;
beta amplitude, and N200 amplitude varied systematically across ERP 5 event-related potential.
Dynamics of concurrent CNS and ANS activity 547

Figure 3. The grand-average waveforms of skin conductance responses ~SCRs! ~A!, electroencephalogram power spectra ~B!, and
event-related potentials ~ERPs! ~C! evoked by targets 1 to 10 and targets 15, 20, 25, 30, 35, and 40. The divisions on the vertical axis
in panels ~A!, ~B!, and ~C! are 0.25 mS, 2 mV 20Hz, and 10 mV, respectively. Each target label marks the position of the zero baseline
for the target trace. In ~C!, the dotted line is the zero baseline based on 200 ms prestimulus data for each target ERP trace. Upward
deflection indicates negativity. The SCR peak amplitude is decreasing ~A! and the alpha and beta band powers are increasing ~B! and
the N200 amplitude at about 200 ms is also increasing ~C! with the advancing sequence of targets. Note in C, the relative sizes of N100
and N200 and that the N200 amplitude above the zero was small for targets 1 to 5, with an increase in N200 amplitude thereafter.

Discussion with ISIs of 20–30 s in nonattend condition, in which responses

usually extinguish after 5–20 min of stimulation and reappear when
The pattern and directions of change of SCR amplitude, alpha and there is a change in any stimulus feature ~Näätänen, 1992, p. 64!.
beta amplitudes, and N200 amplitude across the trial are evident in To date, little data have been published on the characteristics of
Figure 3. SCR amplitude correlated negatively with EEG alpha, SCRs obtained in an oddball paradigm with short ISIs of 1–2 s.
EEG beta, and N200 amplitude ~Table 2!. In addition, the SCR One of the reasons is that scoring overlapping SCRs has been a
amplitude exhibited an exponential decline opposite to the incline difficult task ~Barry et al., 1993; Lyytinen et al., 1992! until re-
of EEG alpha, EEG beta, and N200 ERP over time of target pre- cently ~Lim et al., 1997!. The other is the desire to avoid overlap-
sentations ~Figure 4!. Moreover, their fitted exponential functions ping responses all together ~Roth et al., 1991!.
had a time constant between 1 and 2 min ~Table 1!. It is notable A number of EDA habituation results obtained using ISIs be-
that the amplitude dynamics of these diverse measures of brain tween 15 and 120 s indicate that longer ISIs produce larger re-
function, especially the dynamics of SCR and N200 amplitude, sponses and slower response habituation than shorter ISIs ~Siddle,
should have such a narrow time constant range. Stephenson, et al., 1983!. Responses to attended stimuli of short
SCR as an index of OR exhibits response habituation to un- ISIs can be obtained and scored, and the differences between re-
attended repetitive stimuli of a wide variety of physical features sponses to attended stimuli of short ISI from those obtained using
~for review see Boucsein, 1992; and also Siddle, Stephenson, & a classical habituation paradigm with long ISI has been described
Spinks, 1983!. Response habituation has been studied extensively elsewhere ~Lim et al., 1999!. Briefly, the former exhibit higher
548 C.L. Lim et al.

Table 2. The Correlates of log (SCR)

Number of Correlation
Independent data points coefficient
variable ~n! ~k! p

Alpha 39 20.64 ,.0005

Beta 38 20.61 ,.0005
N200 39 20.50 ,.001

Note: SCR 5 skin conductance response.

response rate, greater response prevalence, and faster rate of de-

cline than the latter. Despite these differences, a key common
feature is the response habituation in both paradigms. The rate of
SCR decrement in this current study appears to be consistent with
the general trend of faster habituation with shorter ISI, and it
perhaps defines a new boundary of response habituation rate. Be-
cause habituation is widely regarded as an elementary form of
learning ~Stephenson & Siddle, 1983!, this boundary may contrib-
ute to the understanding of learning process especially in its early
stages of memory consolidation.
Could the close correspondence of ERP and SCR response
characteristics be due to the sweat gland activity on the scalp being
embedded in the ERP? This explanation is unlikely, because the
N200 onset occurred at about 150 ms and lasted less than 200 ms,
whereas SCR onset in the fingers was about 1,720 ms. The net
difference in onset time on the scalp was more than 900 ms after
allowing for the time difference of 800 ms ~for the head-finger
conduction time difference over a head-hand length difference of
0.8 m for a c fibre conduction velocity of about 1 m0s @Shahani,
Halperin, Boulu, & Cohen, 1984#!. Moreover, the SCR rise time
~peak time to onset time! of about 2,500 ms is an order of mag-
nitude greater than that of N200.
The increment of N200 over trial, in an attend condition, has
not been reported previously. N200 in nonattend conditions has
been found to increase progressively in size from wakefulness
through stage I to stage II sleep, and has been considered to reflect
an arousal effect ~Picton et al., 1974!. This finding can possibly
explain N200 increment with advancing targets across the trial,
perhaps associated with decreasing level of arousal with lapsed
time. However, our paradigm involved active auditory discrimina-
tion and reaction time response in a relatively short test, during
which the tonic skin conductance level was reasonably stable. The
SCL did not correlate with any of the ERP wave peak or with SCR.
This argues against the influence of global tonic variations in
arousal. An ERP peak ~O-wave! increment with trial repetition has
been reported in an unattend condition ~Simons et al., 1987!. We
believe that the N200 increment seen in our study reflects a central
adaptive response to stimulus repetition not directly associated
with general tonic arousal.
Because button press to target detection has been found to
cause substantial slow negative shift in ERPs prior to stimulus
~Starr et al., 1995!, our observed N200 amplitude increment could
potentially be due to differences in negative shifts over trial, namely
reflecting changes in the readiness potential. This possibility is
Figure 4. The temporal variations of actual data ~open! and fitted expo-
unlikely, because the ERP peaks in this study were measured rel-
nential curves ~solid! for: skin conductance response ~SCR! ~A!; electro-
encephalogram alpha ~circle! and beta ~triangle! ~B!; and event-related
ative to the averaged level in the prestimulus 200-ms period, and
potential N200 ~C! with the same advancing target time on the horizontal according to these authors when the slow shifts were removed, the
axis. The fitted curves depict the exponential decline of SCR, and the ERPs associated with count and button press paradigms did not
exponential rise of alpha, beta, and N200, with time. The detailed param- differ ~Starr et al., 1995!. The lack of ERP waveform differences
eters of the fitted exponential time functions are shown in Table 1. between count and button press paradigms in their study also argues
Dynamics of concurrent CNS and ANS activity 549

against the influence of motor-related potentials on endogenous involuntary ~novelty related, unattended! and a voluntary process
ERP. Reaction time correlates more with N200 latency than P300 ~significance-related, attended!. Näätänen ~1992, p. 64! proposes
latency, and increasing reaction times is associated with increasing two components for OR, namely arousal ~nonspecific generalized!
N200 amplitude ~Ritter et al., 1979!. Furthermore, N200 peaks and attentional ~informational!. MMN falls into the arousal cat-
50 ms before EMG response onset, and P300 peaks 100 ms after egory, and apart from the first stimulus during the session, the bulk
EMG responses onset ~Starr et al., 1995!. These findings indicate of our data may be classified into their attentional category. Näätänen
that sensory discrimination decisions have been made around the ~1992, pp. 241–243! argues that in sensory discrimination tasks,
time of N200, and suggest ERP components contributing to N200 N2b is an index of detecting MMN exceeding a threshold. Asso-
play a role in sensory discrimination. ciation of N2b with an integrative detector, accumulating evidence
The close temporal relationship between SCR and N200 ampli- of mismatches to surpass a threshold, has been made by others
tude in this study suggests a link between the central and autonomic ~Loveless, 1986!. An adaptive process might unfold, with iterative
OR. The dynamic nature of the N200 and its smaller amplitude than threshold changes over time, from the initial fuzzy boundary to a
P300 might have contributed to the paucity of description of N200 critical boundary as described by the exponential rise time constant
in the literature. The N200 in this study may contain the bulk of N2b, for N200 ~and SCR! over trial. The iterative process may include
and perhaps part of the positive-going slope of P3a in the N2b–P3a the strengthening of the neuronal representation of the stimulus,
wave complex discussed in the introduction. An alternative view is which occurs during conscious deviance detection as part of the
that N2b does not change in response to repetitive targets and what ongoing sensory discrimination task ~Schröger, 1997!. This dy-
is varying has been P3a, being large in the initial few targets and namic aspect of N2b, evidenced in our data, has not been demon-
habituated later. The summated N2b and P3a waves could have re- strated before. Its temporal variation with SCR response has been
sulted in an apparent reduction of N200 initially, increasing later with too difficult to demonstrate without the SC decomposing tool.
decreasing contribution of P3a. Conceptually, the two scenarios are Finally, the EEG alpha and beta trends across the trial needs to
plausible and the literature seems to favor the notion of an N2b– be differentiated from those of SCR and N200, in that the EEG was
P3a wave complex. Because the positive P3a peak is small, has al- a prestimulus measure and hence not a direct consequence of the
ways been embedded in the positive going slope of P3b ~P300 or target stimulation. The pseudorandomized target presentation in
P3!, and is difficult to measure, the N200 wave peak provides a mea- the paradigm would have minimized the influence of an anticipa-
surable quantity reflecting this complex. The temporal patterns over tory process. The initial lower EEG amplitude or EEG attenuation
trial of both SCR and N200 shared an almost identical time con- ~Figures 3B and 4B!, may be attributed to increased arousal. As the
stant ~Table 1!. This close dynamic relationship between N200 and experiment progressed, the level of arousal decreased, EEG syn-
SCR, sharing identical time constant provides further evidence for chronicity was enhanced and EEG amplitude increased—findings
the N2b–P3a wave complex as a central concomitant of an auto- all largely consistent with previous research ~Davis, 1939; Linds-
nomic electrodermal OR. Further studies are needed to differentiate ley, 1952!. Accordingly, one may argue that the background cor-
the two scenarios. tical activity reflects the general level of arousal, which might
Interpretation of N2b, P3a, or N2b–P3a wave complex, being a influence ~either separately or collectively! N200 and SCR. How-
central OR, leads to the general expectation of a positive relation- ever, in this study, the SCL, which is an index of arousal ~Bohlin,
ship between N200 and SCR. But N200 was found to increase with 1971; Boucsein, 1992!, did not show any trend over the trial.
SCR decrement in this study. This apparent directional discrepancy Therefore changes in EEG cannot be solely due to lowering arousal.
between the empirical data and the general expectation has also On the other hand, EEG desynchronization has been positively
been observed in a nonattend condition ~Simons et al., 1987!. This linked to activation and attentional demands ~Boiten, Sergeant &
discrepancy may be explained as follows: The progressive incre- Geuze; 1992; Dujardin et al., 1993; van Winsum, Sergeant, &
ment of N200 responses to the initial six or seven targets as seen in Geuze, 1984!, and it is thought to be due to cortical excitation
our study, would have been lost in a conventional average because through disinhibition of thalamic interneurons ~Pfurtscheller &
the grand-averaged ERPs would have been dominated by the later, Aranibar, 1977!. Furthermore, coexistence of both ERD in discrete
larger responses, leading to a final averaged tracing with a sizeable cortical regions engaged in a task, and of event-related synchro-
N200. The initial increment is lost in the process. A corollary to the nization in regions not engaged in the task ~Pfurtscheller, 1992!,
explanation is that a fully developed N200 or a diminished P3a rep- argues more for changes associated with information processing
resents an habituated central OR and the initial incremental change than for global changes in arousal.
of N200 reflects central OR habituation in progress. In summary, it is not surprising that there are modulations
If this explanation is accepted, the network substrate underlying underlying central and autonomic functions. What was surprising
incremental dynamic changes of N200 may also be linked to pro- in this study was the similarity in temporal patterns displayed by
gressive consolidation of the neuronal model according to Sokolov’s different measures of brain function, which had time constants of
comparator theory ~1960, 1963, 1975, 1990!. The substrate for the 1–2 min, and the selective covariation of the N200 component with
neuronal model has been associated with cortical regions and the SCR. This is the first demonstration of the selective and dynamic
hippocampus ~O’Donnell et al., 1993; Salisbury, O’Donnell, Mc- relationships between SCR and N200. The 1–2-min time constant
Carley, Shenton, & Benavage, 1994; Sokolov, 1975; Vinogradova, provides a clue to the conjoint processes that underlie these central
1975!. Sokolov ~1990! argues that there are two forms of OR, an and autonomic adaptive functions.


Barry, R. J., Feldmann, S., Gordon, E., Cocker, K. I., & Rennie, C. ~1993!. Basar, E., Basar-Eroglu, C., Rosen, B., & Schütt, A. ~1984!. A new ap-
Elicitation and habituation of the electrodermal response in a short proach to endogenous event-related potentials in man: Relation be-
interstimulus interval paradigm. International Journal of Psychophys- tween EEG and P300-wave. International Journal of Neuroscience, 24,
iology, 15, 247–253. 1–21.
550 C.L. Lim et al.

Bloch, V. ~1993!. On the centennial of the discovery of electrodermal examination in electroencephalography. Electroencephalography and
activity. In J.-C. Roy, W. Boucsein, D. Fowles, & J. Gruzelier ~Eds.!, Clinical Neurophysiology, 10, 370–375.
Progress in electrodermal research ~pp. 1–5!. New York: Plenum Press. Kellaway, P. ~1990!. An orderly approach to visual analysis: Characteristics
Bohlin, G. ~1971!. Monotonous stimulation, sleep onset and habituation of of the normal EEG of adults and children. In D. D. Daly & T. A. Pedley
the orienting reaction. Electroencephalography and Clinical Neuro- ~Eds.!, Current practice of clinical electroencephalography ~pp. 139–
physiology, 31, 593– 601. 199!. New York: Raven Press.
Boiten, F., Sergeant, J., & Geuze, R. ~1992!. Event-related desynchroniza- Lim, C. L., Barry, R. J., Gordon, E., Sawant, A., Rennie, C., & Yiannikas,
tion: The effects of energetic and computational demands. Electro- C. ~1996!. The relationship between quantified EEG and skin conduc-
encephalography and Clinical Neurophysiology, 82, 302–309. tance level. International Journal of Psychophysiology, 21, 151–162.
Boucsein, W. ~1992!. Electrodermal activity. New York: Plenum Press. Lim, C. L., Gordon, E., Harris, A., Bahramali, H., Li, W. M., Manor, B., &
Brazier, M. A. B. ~1961!. A history of the electrical activity of the brain. Rennie, C. ~1999!. Electrodermal activity in schizophrenia: A quanti-
London: Pitman. tative study using a short interstimulus paradigm. Biological Psychia-
Cobb, W. A. ~1963!. The normal adult EEG. In D. Hill & G. Parr ~Eds.!, try, 45, 127–135.
Electroencephalography—A symposium on its various aspects ~pp. 232– Lim C. L., Rennie, C., Barry, R. J., Bahramali, H., Lazzaro, I., Manor, B.,
249!. London: MacDonald. & Gordon E. ~1997!. Decomposing skin conductance into tonic and
Damasio, A. R. ~1994!. Descartes’ error: Emotion, reason and human phasic components. International Journal of Psychophysiology, 25,
brain. London: Picador. 97–107.
Davies, D. R., & Krkovic, A. ~1964!. Skin conductance, alpha-activity, and Lindsley, D. B. ~1952!. Psychological phenomena and the electroenceph-
vigilance. American Journal of Psychology, 78, 304–306. alogram. Electroencephalography and Clinical Neurophysiology, 4, 443–
Davis, P. A. ~1939!. Effects acoustic stimuli on the waking brain. Journal 456.
of Neurophysiology, 2, 494– 499. Lopes Da Silva, F. ~1991!. Neural mechanisms underlying brain waves:
Dawson, M. E., Schell, A. M., & Filion, D. L. ~1990!. The electrodermal From neuronal membranes to networks. Electroencephalography and
system. In J. T. Cacioppo, & L. G. Tassinary ~Eds.!, Principles of Clinical Neurophysiology, 79, 81–93.
psychophysiology: Physical, social, and inferential elements ~pp. 295– Loveless, N. ~1983!. The orienting response and evoked potentials in man.
324!. Cambridge, UK: Cambridge Press. In D. Siddle ~Ed.!, Orienting and habituation: Perspective in human
Donchin, E. ~1979!. Event-related brain potentials: A tool in the study of research ~pp. 109–182!. New York: Wiley.
human information processing. In H. Begleiter ~Ed.!, Evoked potentials Loveless, N. E. ~1986!. Potentials evoked by temporal deviance. Biological
and behavior ~pp. 13–75!. New York: Plenum Press. Psychology, 22, 149–167.
Donchin, E. ~1981!. Surprise! . . . Surprise? Psychophysiology, 18, 493– Lykken, D. T., & Venables, P. H. ~1971!. Direct measurement of skin
513. conductance: A proposal for standardisation. Psychophysiology, 8, 656–
Donchin, E., Heffley, E., Hillyard, S. A., Loveless, N., Maltzman, I., Öhman, 672.
A., Rösler, F., Ruchkin, D., & Siddle, D. ~1984!. Cognition and event- Lyytinen, H., Blomberg, A. P., & Näätänen, R. ~1992!. Event-related po-
related potentials II: The orienting reflex and P300. Annals of the New tentials and autonomic responses to a change in unattended auditory
York Academy of Sciences, 425, 39–57. stimuli. Psychophysiology, 29, 523–534.
Dujardin, K., Derambure, P., Defebvre, L., Bourriez, J. L., Jacquesson, Lyytinen, H., & Näätänen, R. ~1987!. Autonomic and ERP responses to
J. M., & Guieu, J. D. ~1993!. Evaluation of event-related desynchroni- deviat stimuli: analysis of covariation. In R. Johnson, Jr., J. W. Rohr-
zation ~ERD! during a recognition task: Effect of attention. Electro- baugh, & R. Parasuraman ~Eds.!, Current trends in event-related brain
encephalography and Clinical Neurophysiology, 86, 353–356. potentials research ~EEG Suppl. 40! ~pp. 108–117!. Amsterdam: Elsevier.
Edelberg, R. ~1993!. Electrodermal mechanisms: A critique of the two- Montplaisir, J. Y. ~1975!. Cholinergic mechanisms involved in cortical
effector hypothesis and a proposed replacement. In J.-C. Roy, W. Bouc- activation during arousal. Electroencephalography and Clinical Neuro-
sein, D. Fowles, & J. Gruzelier ~Eds.!, Progress in electrodermal research physiology, 38, 263–272.
~pp. 7–29!. New York: Plenum Press. Moruzzi, G., & Magoun, H. W. ~1949!. Brain stem reticular formation and
Ford, J. M., Roth, W. T., Dirks, S. J., & Kopell, B. S. ~1973!. Evoked activation of the EEG. Electroencephalography and Clinical Neuro-
potentials correlates of signal recognition between and within modal- physiology, 1, 455– 473.
ities. Science, 181, 465– 466. Näätänen, R. ~1982!. Processing negativity: An evoked potential reflection
Ford, J. M., Roth, W. T., & Kopell, B. S. ~1976!. Auditory evoked potentials of selective attention. Psychological Bulletin, 92, 605– 640.
to unpredictable shifts in pitch. Psychophysiology, 13, 32–39. Näätänen, R. ~1992!. Attention and brain function. Hillsdale, NJ: Erlbaum.
Fowles, D. C. ~1986!. The eccrine system and electrodermal activity. In Näätänen, R., & Gaillard, A. W. K. ~1983!. The orienting reflex and the N2
M. G. H. Coles, E. Donchin, & S. W. Proges ~Eds.!, Psychophysiology: deflection of the event-related potential ~ERP!. In A. W. K. Gaillard &
System, processes, applications ~pp. 51–96!. New York: Guilford Press. W. Ritter ~Eds.!, Tutorials in ERP research: Endogenous components
Fowles, D. C., Christie, M. J., Edelberg, R., Gringes, W. W., Lykken, D. T., ~pp. 119–141!. Amsterdam: North-Holland.
& Venables, P. H. ~1981!. Publication recommendations for electroder- Näätänen, R., & Picton, T. ~1987!. The N1 wave of the human electric and
mal measurements. Psychophysiology, 18, 232–239. magnetic response to sound: A review and analysis of the component
Gloor, P. ~1969!. Hans Berger on electroencephalogram of man. Electro- structure. Psychophysiology, 24, 375– 425.
encephalography and Clinical Neurophysiology ~Suppl. 28!, 1–36. Niedermeyer, E. ~1987!. The normal EEG of the waking adult. In E. Nieder-
Graham, F. K. ~1973!. Habituation and dishabituation of responses inner- meyer & F. Lopes da Silva ~Eds.!, Electroencephalography—Basic
vated by the autonomic nervous system. In H. V. S. Peeke & M. J. Herz principles, clinical application and related fields ~pp. 97–117!. Balti-
~Eds.!, Habituation: Behavioral studies, Volume 1 ~pp. 163–218!. New more, MD: Urban & Schwarzenberg.
York: Academic Press. O’Donnell, B. F., Shenton, M. E., McCarley, R. W., Faux, S. F., Smith,
Gratton, G., Coles, M. G., & Donchin, E. ~1983!. A new method for off-line R. S., Salisbury, D. F., Nester, P. G., Pollak, S. D., Kikinis, R., & Jolesz,
removal of ocular artifact. Electroencephalography and Clinical Neuro- F. A. ~1993!. The auditory N2 component in schizophrenia: Relation-
physiology, 55, 468– 484. ship to MRI temporal lobe gray matter and to other ERP abnormalities.
Halgren, E., & Marinkovic, K. ~1994!. Neurophysiological networks inte- Biological Psychiatry, 34, 26– 40.
grating human emotions. In M. S. Gezzaniga ~Ed.!, The cognitive neuro- Pavlov, I. P. ~1927!. Conditioned reflexes. New York: Oxford University
sciences ~pp. 1137–1151!. Cambridge, MA: MIT Press. Press.
Halliday, A. M. ~1980!. Evoked brain potentials: How far have we come Perrault, N., & Picton, T. W. ~1984!. Event-lated potentials recorded from
since 1875? In C. Barber ~Ed.!, Evoked potentials ~pp. 3–18!. Lan- the scalp and nasopharynx. II. N2, P3 and slow wave. Electroenceph-
caster, PA: MTP Press. alography and Clinical Neurophysiology, 59, 261–278.
Hillyard, S. A., Hink, R. F., Schwent, V. L., & Picton, T. W. ~1973!. Pfurtscheller, G. ~1992!. Event-related cortical synchronization ~ERS!: An
Electrical signs of selective attention in the human brain. Science, 182, electrophysiological correlate of cortical areas at rest. Electroenceph-
177–180. alography and Clinical Neurophysiology, 83, 62– 69.
Intriligator, J., & Polich, J. ~1994!. On the relationship between background Pfurtscheller, G., & Aranibar, A. ~1977!. Event-related cortical desynchro-
EEG and the P300 event-related potential. Biological Psychology, 37, nization detected by power measurements of scalp EEG. Electroenceph-
207–218. alography and Clinical Neurophysiology, 42, 817–826.
Jasper, H. H. ~1958!. Report of the committee on methods of clinical Picton, T. W., Hillyard, S. A., Krauz, H. I., & Galambos, R. ~1974!. Human
Dynamics of concurrent CNS and ANS activity 551

auditory evoked potential. I. Evaluation of components. Electroenceph- M. A. B. Brazier ~Ed.!, The central nervous system and behavior
alography and Clinical Neurophysiology, 36, 179–190. ~pp. 187–275!. New York: Josiah Macey Jr. Foundation.
Press, W. H., Flannery, B. P., Teukolsky, S. A., & Vetterling, W. T. ~1992!. Sokolov, E. N. ~1963!. Perception and the conditioned reflex. Oxford, UK:
Modelling of data. In Numerical recipes in C: The art of scientific Pergamon Press.
computing ~2nd ed., pp. 656– 699!. Cambridge, UK: Cambridge Uni- Sokolov, E. N. ~1975!. The neuronal mechanisms of the orienting reflex. In
versity Press. E. N. Sokolov & O. S. Vinogradova ~Eds.!, Neuronal mechanisms of
Ritter, W., Simon, R., Vaughan, H. G., & Friedman, D. ~1979!. A brain the orienting reflex ~pp. 217–235!. Hillsdale, NJ: Erlbaum.
event related to the making of a sensory discrimination. Science, 203, Sokolov, E. N. ~1990!. The orienting response, and future directions of its
1358–1361. development. Pavlovian Journal of Biological Sciences. 25, 142–150.
Ritter, W., Vaughan, H. G., & Costa, L. D. ~1968!. Orienting and habitu- Squires, N. L., Squires, K. C., & Hillyard, S. A. ~1975!. Two varieties of
ation to auditory stimuli: A study of short term changes in average long-latency positive waves evoked by unpredictable auditory stimuli
evoked responses. Electroencephalography and Clinical Neurophysi- in man. Electroencephalography and Clinical Neurophysiology, 39,
ology, 25, 550–556. 387– 401.
Roth, W. T., Goodale, J., & Pfefferbaum, A. ~1991!. Auditory event-related Starr, A., Sandroni, P., & Michalewski, H. J. ~1995!. Readiness to respond
potentials and electrodermal activity in medicated and unmedicated in a target detection task: Pre- and post-stimulus event-related poten-
schizophrenics. Biological Psychiatry, 29, 585–599. tials in normal subjects. Electroencephalography and Clinical Neuro-
Roy, J.-C., Sequeira, H., & Delerm, B. ~1993!. Neural control of electro- physiology, 96, 76–92.
dermal activity: Spinal and reticular mechanisms. In J.-C. Roy, W. Stephenson, D., & Siddle, D. ~1983!. Theory of habituation. In D. Siddle
Boucsein, D. Fowles, & J. Gruzelier ~Eds.!, Progress in electrodermal ~Ed.!, Orienting and habituation: Perspective in human research
research ~pp. 73–92!. New York: Plenum Press. ~pp. 183–236!. New York: Wiley.
Salisbury, D. F., O’Donnell, B. F., McCarley, R. W., Shenton, M. E., & Steriade, M., Gloor, P., Llinàs, R. R., Lopes da Silva, F. H., & Mesulam,
Benavage, A. ~1994!. The N2 event-related potential reflects attention M. M. ~1990!. Basic mechanisms of cerebral rhythmic activities. Elec-
deficit in schizophrenia. Biological Psychology, 39, 1–13. troencephalography and Clinical Neurophysiology, 76, 481–508.
Schröger, E. ~1997!. On the detection of auditory deviations: A pre- Sutton, S., Baren, M., Zubin, J., & John, E. R. ~1965!. Evoked potentials
attentive activation model. Psychophysiology, 34, 245–257. correlates of stimulus uncertainty. Science, 150, 1187–1188.
Schwent, V. L., Hillyard, S. A., & Galambos, R. ~1976!. Selective attention van Winsum, W., Sergeant, J., & Geuze, R. ~1984!. The functional sig-
and the auditory vertex potentials, I: Effects of stimulus delivery rate. nificance of event-related desynchronization of alpha rhythm in at-
Electroencephalography and Clinical Neurophysiology, 40, 604– 614. tentional and activating tasks. Electroencephalography and Clinical
Shahani, B. T., Halperin, J. J., Boulu, P., & Cohen, J. ~1984!. Sympathetic Neurophysiology, 58, 519–524.
skin response—A method of assessing unmyelinated axon dysfunction Verleger, R. ~1988!. Event-related potentials and cognition: A critique of
in peripheral neuropathies. Journal of Neurology Neurosurgery and the context updating hypothesis and an alternative interpretation of P3.
Psychiatry, 47, 536–542. Behavioral and Brain Science, 11, 343– 427.
Siddle, D. A. T., Remington, B., Kuiack, M., & Haines, E. ~1983!. Stimulus Vinogradova, O. S. ~1975!. The hypocampus and the orienting reflex. In
omission and dishabituation of the skin conductance response. Psycho- E. N. Sokolov & O. S. Vinogradova ~Eds.!, Neuronal mechanisms of
physiology, 20, 136–145. the orienting reflex ~pp. 128–154!. Hillsdale, NJ: Erlbaum.
Siddle, D. A. T., Smith, D. G., & Marcer, D. ~1974!. Effects of repetitive Voronin, L. G., Bonfitto, M., & Vasilieva, V. M. ~1975!. The interrelation
stimulation and reduced sensory input on cortical alertness. Biological of orienting reaction and conditioned reflex to time in man. In E. N.
Psychology, 1, 181–188. Sokolov & O. S. Vinogradova ~Eds.!, Neuronal mechanisms of the
Siddle, D., Stephenson, D., & Spinks, J. A. ~1983!. Elicitation and habit- orienting reflex ~pp. 252–263!. Hillsdale, NJ: Erlbaum.
uation of the orienting response. In D. Siddle ~Ed.!, Orienting and Wallin, B. G. ~1981!. Sympathetic nerve activity underlying electrodermal
habituation: Perspective in human research ~pp. 109–182!. New York: and cardiovascular reactions in man. Psychophysiology 18, 470– 476.
Simons, R. F., Rockstroh, B., Elbert, T., Fiorito, E., Lutzenberger, W., &
Birbaumer, N. ~1987!. Evocation and habituation of autonomic and
event-related potential responses in a nonsignal environment. Journal
of Psychophysiology, 1, 45–59.
Sokolov, E. N. ~1960!. Neuronal models and the orienting response. In ~Received August 19, 1997; Accepted October 21, 1998!