42 Elec troencephalography and clinical Neurophysiology , 91 (1994) 42-53 © 1994 Elsevier Science Ireland Ltd. 0013-4694/94/$07.00

EEG 93163

Dynamic properties of human visual evoked and omitted stimulus potentials

Theodore H. Bullock *, Sacit Karamiirsel, Jerzy Z. Achimowicz, Michael C. McClune and Canan Ba§ar-Eroglu * *

Neurobiology Unit, Scripps Institution of Oceanography, and Department of Neurosciences, School of Medicine, University of California, San Diego, 9500 Gilman Dr., Dept. 0201, La Jolla, CA 92093-0201 (USA)

(Accepted for publication: 21 January 1994)

Summary Visual evoked potentials (VEPs) and omitted stimulus potentials (OSPs) are re-examined in scalp recordings from 19 healthy subjects. The principal finding is a distinction in form, latency and properties between OSPs in the conditioning stimulus range < 2 Hz, used in previous human studies, and those in the range > 5 Hz, used in previous studies of selected elasmobranchs, teleost fish and reptiles. We cannot find OSPs between 2 and 5 Hz. The high frequency ("fast," ca. 6- > 40 Hz) and the low frequency ("slow," ca. 0.3-1.6 Hz) OSPs have different forms and latencies but both tend to a constant latency after the omission, over their frequency ranges, suggesting a temporally specific expectation. Fast OSPs (typically N120, P170-230 and later components including induced rhythms at 10-13 Hz) resemble an OFF effect, and require fixation but not attention to the interstimulus interval. Slow OSPs (usually P500-1100) require attention but not fixation; they are multimodal, unlike the fast OSPs. Based on cited data from fish and reptiles, fast OSPs probably arise in the retina, to be modified at each subsequent level. We have no evidence on the origin of slow OSPs. In both ranges not only large, diffuse flashes, but weak, virtual point sources (colored LEDs) meters away suffice. They are difficult to habituate. Both require very short conditioning periods.

The transition from the single, rested VEP to the steady state response (SSR) at different frequencies is described. Around 8-15 Hz in most subjects larger SSRs suggest a resonance. Alternation between large and small SSR amplitude occurs around 4 Hz in some subjects and conditions of attention, and correlates with an illusion that the flash frequency is 2 Hz or is irregular. Jitter of the conditioning intervals greatly reduces the slow OSP but only slightly affects the fast OSP. Differences between scalp loci are described.

Key words: Cognitive potential; Event-related potential; Steady state response; Expectation; Attention

Event-related potentials (ERPs) to missing stimuli have been known in human subjects for many years (Rusinov 1960; Barlow et al. 1965; Sutton et al. 1965, 1967; Klinke et al. 1968; Picton et al. 1974; Squires et al. 1975, 1977; Renault and Les6vre 1977, 1978; Kerkhof 1978; Courchesne 1979; Ritter et al. 1979; Ruchkin and Sutton 1979a,b; McCallum 1980; Ford and Hillyard 1981; Takasaka 1985; Hillyard and Picton 1987; Alain et al. 1989; Mangun et al. 1993). However, their de- pendence upon the interstimulus interval (ISI) of the series prior to the omitted stimulus, as well as upon the duration and regularity of the series have received little attention. These dynamic properties became important when it was discovered that ERPs to omitted stimuli are prominent in non-mammalian vertebrates. They

* Corresponding author. Tel.: 619/534-3636; Fax: 619/534-3919; E-mail: tbullock@ucsd.edu. ** Present address: Physiological Institute, Medical University, Liibeck, Germany.

are already well developed in the midbrain tectum and even in the retina after conditioning series of light flashes in rays and turtles (Bullock et al. 1990; Prechtl and Bullock 1992, 1994) as well as in primary sensory nuclei of the medulla after physiological electrosensory stimuli in rays (Bullock et al. 1993). In these studies the most effective conditioning frequencies were found to be in a range (> 2 Hz) not heretofore explored in human subjects. The latency of the omitted stimulus potential (OSP) in these animals is much more pre- cisely constant after the due-time of the omission, over a range of conditioning ISis, than has been noted in humans, suggestive of a temporally specific expecta- tion. The effect of irregularity of ISI (hereafter "jitter") is minimal in these species of fish and reptiles in their effective frequency range for OSPs, whereas the lim- ited information on humans (Takasaka 1985) indicates that jittered stimulus series give no average OSP.

The aims of the present study are to examine the neglected range > 2 Hz in normal human subjects, to measure the latency of the OSP after different condi-

SSDI 0013-4694(94)00025-G

HUMAN VEP AND OSP 43

tioning frequencies, to report a surprisingly short mini- mum conditioning series and a more careful test of the effect of irregularity of ISis. Relevant to these ques- tions also is better information on the transition of visual evoked potentials (VEPs) from the rested re- sponse after a single flash to the steady state response (SSR) at different ISis. We report in addition on amplitude alternation of successive VEPs under cer- tain conditions, including attention, on the effects of binocular stimulation, cross-modal stimulation, and dif- ferences in dependence on attention in the "high" (> 2 Hz) and "low" ( < 2 Hz) frequency ranges. The occurrence under some conditions of induced rhythms, fluctuations in sensitivity and regional differences in response are noted.

Materials and methods

Subjects

Under protocols from the human subjects commit- tees of the Universities of California and of Liibeck, 19 normal volunteers, clinically normal, high functioning adults, aged 30-75 (4 females), mostly scientists famil- iar with the project, were included in this study. No differences in results appeared to be correlated with age or gender. During the recording session, subjects sat in a chair in a dimly illuminated room, in a relaxed position. They were usually asked not to move or speak, or to blink too much, and to look at a fixation point on the flash probe. Subjects were instructed to try to predict the time of the next flash or, in the case of low frequency stimulation with single omissions, to count the omissions. In the case of low frequency stimulation with jittered ISis, they were instructed to try to predict the time of the next flash. In the case of high frequency (> 5 Hz) stimulation, usually given in short trains to permit seeing the whole OSP without subsequent VEPs, they were given no instruction. Above ca. 4 Hz, subjects report it soon becomes impos- sible to keep track of each flash or its absence. Short rest periods were provided between recording sessions. The number of recording sessions in a day together amounted to 2.5-3.5 h.

Recording

E EG data were taken via Ag-AgC1 disk electrodes or gold electrodes (Grass). The active electrodes were placed according to the international 10/20 system at Cz, Fz, Oz, O1, 02, P3, and P4, or at some of them. The reference electrode was on the left ear and grounded electrodes on both ears. E E G signals were amplified via Grass amplifiers with 60 Hz notch filters

and the output sent to steep analog bandpass filters (Rockland, model 452) set variously, from 0.3-100 Hz to 0.1-50 Hz. E E G data were collected in a PDP 11/73 computer which also controlled the stimuli. The E E G data were monitored by oscilloscopes and a Nico- let 1170 computer during the recording sessions. Off- line analysis was performed on the PDP 11/73.

Stimulation

Visual stimuli (8/xsec pulses of bright flashes) were generated with a Grass PS 2 photo-stimulator or a glow modulator driven by the PDP 11/73 computer a n d / o r a PulseMaster A300 digital stimulator. Flash intensity with the Grass strobe light was no. 1 or no. 2 for all paradigms and the flash unit was covered with paper and placed 0.5-1 m from the subject. Small green LEDs were also used for visual stimulation. Auditory stimuli were given as binaural clicks via a stereo head- phone.

Stimulus paradigms

Single flash VEPs were taken with at least 4 sec intervals (ISI) and the effects of the stimulus intensity and the ambient illumination were investigated in a few subjects. The SSR and the OSPs to visual stimula- tion were recorded at 0.3, 0.5, 1, 2, 4, 5, 6.7, 10, 15, 20, 30 and 40 Hz flash frequencies. Two stimulus regimes were used. In the train paradigm 2-30 sec trains (con- ditioning periods) were used after 2-30 sec rest peri- ods. In the continuous series paradigm 1 stimulus was omitted every 5-10 flashes, pseudorandomly. In a vari- ant of the continuous series, the OSP to omission of every third stimulus was compared with those after omission of the fourth, fifth or sixth stimulus in a regular series.

Cross-modality effects were found in the equiva- lence, for certain frequencies, of a terminal stimulus of a different modality (click ending a flash train or flash ending a click train).

To test for the effects of irregularity of intervals the ISis were jittered in a low and a high percentage of the mean, with the condition that the last interval was equal to the mean.

Experiments were done in two different periods in Liibeck and an extended period in La Jolla, with differ- ent set-ups and overlapping personnel. Most of the techniques and procedures were quite standard or dif- fered in ways that proved quite insignificant, according to the results, for example placement of electrodes, filter settings and methods of rejecting sweeps. No doubt, however, there were unintentional differences that might have affected results, for example in instruc- tions to subjects and state of the subjects. Our judg-

44 T.H. BULLOCK ET AL.

ment is that these were no greater or more systematic between laboratories than within each laboratory. Un- controlled differences between runs or subjects or months surely occurred in each locale. The extremes of the variance of results under the same stimulus regime could occasionally be found in the same subject, some months apart, but successive sessions on the same day or different days, up to months apart gave quite the same results in each subject in > 95% of such compar- isons. In this basically descriptive sampling of human responses we have not undertaken a quantitative study of the variance of selected measures but report the ranges of most measures qualitatively.

Analysis"

The average EP, SSR and OSP responses were obtained from up to 75-100 single sweeps. Selective averaging was used to eliminate movement and eye blink artifacts, i.e., sweeps judged by eye to contain such artifacts were rejected. Steep digital low-pass fil- ters removed high frequency artifacts and E E G compo- nents in some experiments.

Results

The principal surprising result is the distinction be- tween OSPs in high and low frequency conditioning stimulus ranges (see Effects of stimulus frequency on VEP and OSP). Before detailing this finding, we will mention some of the stimulus parameters that can influence the form, amplitude or latency of OSPs. This will involve some new and unexpected dynamic aspects of human scalp VEPs which are the responses to the background train of flashes that must be distinguished from those attributable to omission.

Effects of temporal parameters on OSPs and preceding PEPs

Prolonged stimulation does not substantially influ- ence the measures reported here; it does not habituate or fatigue. In repeated records during 1 h or more of nearly continuous stimulation it is usual to see no changes in the VEPs in either the low flash rate ( < 2 Hz) or high rate ( > 2 Hz) range; each frequency reaches an SSR after a short transition period and maintains this SSR. This observation is not from examining single sweeps but from comparing averages of 50 sweeps; we cannot say what changes may have occurred within that time. The OSPs to low and high rate stimulation like- wise do not change with time unless the subject loses attention, which has to be of a different sort for the two ranges (see Effect of the type of attention on OSP, below).

OSPs to the omission of a single stimulus during a sustained train are similar in morphology to those after the end of a train. They may look more complicated in the former paradigm due to the superimposition of the next VEPs and their transition to an SSR in the high rate range (Fig. 1). At slow rates OSPs to the omission of a single stimulus and to the end of a train are quite the same in form.

Train duration, stimulus number and interval between trains

Each of these parameters can exert some influence on the shape and amplitude of VEPs and OSPs but the effects are not large. In the high frequency range, tests from 5 to 40 Hz show that a 2 sec conditioning period gives OSPs as large as does a 30 sec period and the same is true after 2 sec and 30 sec resting periods following either the long or the short conditioning. We have not studied the beginning of the train at these frequencies in human scalp recordings to see the tran- sition from the first, rested VEP to the SSR but this is reported for the elasmobranch in Karamfirsel and Bul- lock (1994), where it is shown that much less than 2 sec is required for an OSP and only a very few ISis.

In the low frequency range, tests from 0.3 to 2 Hz show that the VEP comes to its final form and ampli- tude after the second stimulus. Amplitude, latency and shapes of the third, fourth and fifth VEPs are exactly the same as the second one, if we examine averages of trains. This was true in each of the 6 subjects examined for this point but we expect to find exceptions since the VEP is so variable in form and amplitude among subjects. We did not study this point quantitatively or at many scalp loci. In the low frequency range OSPs also show a low minimum requirement for preceding ISis; the amplitude, latency and shape of the OSPs to the omission of the third, fourth, fifth and later stimuli look the same. At 0.3, 0.5 and 1 Hz, 2 stimuli are enough, i.e., 1 ISI, to elicit an end of train wave that looks just like a true OSP (Fig. 4A). It is important, however, to point out that these comparisons are made on averages, so that the system has seen a pat terned series of stimuli and omissions. According to the con- clusions from similar experiments on elasmobranch fish (Karamiirsel and Bullock 1994), we can expect that, if the SSR has developed and stabilized, so that the next sequence can be predicted, the conditioning period of that duration is long enough to show a good OSP. This implies that the response to the omission of every third stimulus ( = omission of third, sixth, ninth, etc.) is gen- erated by a mechanism similar to that for omissions of every fourth or fifth or sixth, in this low frequency range.

Effects of stimulus frequency on VEP and OSP No OSP has been observed in any of our subjects to

2 -5 Hz flash stimulation; the VEPs are not followed by

H U M A N VEP A N D OSP 45

an OFF response. Good OSPs are obtained below 2 Hz (1, 0.5 and 0.3 Hz), and above 5 Hz (6.7, 8 . . . . . 40 Hz). A transition zone between 1 and 6.7 Hz separates OSPs of different characteristics for slow and fast stimulation rates. Accordingly, we describe 2 types of OSP in humans (Figs. 1, 3 and 4). Those following fast and slow conditioning stimulus trains we will call fast and slow OSPs.

High stimulus rate (> 2 Hz). The amplitude of the OSP becomes greater with higher conditioning rates in the range of 5-15 Hz and then stays unchanged or in some subjects declines. The SSR becomes smaller with higher frequency and disappears above 20 Hz in most subjects. The lowest frequency at which the SSR disap- pears, even in an average, is the LFF or lowermost fusion frequency by the evoked potential criterion. It is not the same as the subjective flicker fusion frequency, nor is it a fixed value but it depends on intensity, ambient light, peripheral vs. central vision and other factors. It can be as low as 20 Hz - in the same range as we have reported for many of the slower fish species,

whereas other fish species are much higher (Bullock et al. 1990).

Around 8-15 Hz, in some loci in most of the sub- jects, the SSR is much larger than expected from the amplitude at lower rates, suggesting a resonance (Fig. 1A, 15 Hz; Fig. 3, Oz, 10-15 Hz).

The latency of the OSP after the next due-time is quite constant in the stimulus frequency range of 10-15 to 40 Hz, in the same subject. Since below 10 Hz OSPs are small, longer, more complicated, superimposed on the SSR and sometimes with oscillatory components, it is difficult to determine their latencies with accuracy. Variability is considerable between subjects but it is not amenable to quantitative comparison; like the VEP, the form, initial polarity and number of peaks as well as amplitude vary between subjects in scalp recordings, possibly due to differences in the geometry of the striate convolutions. In most subjects, the OSP appears as an N120-P170 and P200-P250. Later components are often seen, at least to 500-700 msec, but are even more variable.

A First Omission a~er Train

Locus 0 z

P220

N120 SSR I

' 1 P230

Locus C, I

_ _ _ a ~

I

I

" i I

~ j

8 ~

' ' i ' ' ' i ' ' ' i ' ' ' i

0 . 4 0 . 6 0 . 8 1 - Seconds

B Single Missing Stimulus

10Hz ~Y ~ ~ / ~ i W ~ / W v V ~ /

N130 J SSR

Locus C= I

lOHz ~

P250

8

0 i ' ' , i ', , , i ' ' ' i ' ' ' i , ' ' i

0 0 . 2 0 . 4 0 . 6 O.B 1 - 8 Seconds

30

Fig. 1. Comparison of (A the end of a train and (B) the omission of 1 st imulus in a high frequency range of stimuli (10-30 Hz), in 2 loci (upper, Oz; lower, Cz), aligned by the due-t ime of the first omitted stimulus. The main peaks of the OSP - some are marked with their polarity and latency after the time of omission - are almost the same in the two paradigms but look more complicated in B due to the superimposition of the

slightly rested VEP and the SSR to stimuli after the omission. Averages of 100.

46 T.H. BULLOCK ET AL.

Alternation of successive responses during an SSR between large and small VEP amplitudes ("1 in 2") is observed in some subjects (in 6 out of 9 where it was looked for; 4 of these showed it well), at certain stimu- lus rates, commonly at 4 Hz (Fig. 2). The SSR follows the flashes 1 : 1 in most subjects, even at high frequen- cies, up to nearly 20 Hz. Every other VEP is larger under special conditions, however, including no eye blinking during the whole recording session, careful fixation to one point on the flash probe held within 10-15 cm from the face and possibly other, unrecog- nized conditions. In that state there is an illusion; subjects report that the 4 Hz flash frequency is 2 Hz. This seems important, because it is a strong indication of the effect of attention on VEPs. Alternations in SSR amplitude are not often seen, as they are in rays (Karamiirsel and Bullock 1994), perhaps because scalp recording averages over too much cortex or because it requires averaging over many trials and alternation is lost in averaging if it does not always begin with the same phase or is not perfectly consistent during the sampling period. We usually need to average at least 30 sweeps of 1 sec each to see a clear response, and hence to maintain the same state of close attention unchanged for at least 30 sec. Fig. 3 shows some 1 in 2 following at 4 Hz in locus Pz.

~vo t- O . 5

-2 Seconds

Fig. 2. Alternation in the ampli tude of successive VEPs and a correlated illusion. Upper trace is the SSR to 4 Hz stimulation, under conditions that cause 1:1 following at constant amplitude - moderate intensity (Grass Photostimulator, intensity no. 1), diffused flash, ca. 1 m from the subject. Vertex derivation. Each trace is the average of 128 sweeps. Second trace: 4 Hz but a special state of attention changes the SSR following to 1 in 2 (alternately smaller and larger VEPs); this must be quite regular to be preserved in such a long average. At this time, the subject reports that the flash rate appears to be 2 Hz. Often when the apparent rate is higher than 3 Hz, a certain kind of at tention causes the illusion that it is irregular. Sometimes the difference between alternate ampli tudes is small and it may be evident only in one or some of the leads. Possibly the regularity is imperfect and averaging smears the alternation. A correlate is that, subjectively, it is difficult to hold the illusion for long. Averages of 74 sweeps; amplifier bandpass 0.3-55 Hz. The st imulus markers are artificially prolonged for visibility; the time of

the brief flash is the onset of the mark.

OSP alternation is not usually observed in averaged records at higher rates; this does not mean it has not occurred, for the reasons just stated. At 20, 30 and 40 Hz, however, the SSR sometimes has a repeating pat- tern that suggests a larger VEP about every seventh flash.

Low stimulus rate (< 2 Hz). VEP amplitude gets smaller with higher stimulus rates. A late component (N250-350) which is conspicuous in the rested VEP (0.1 or 0.25 Hz) is more sensitive than earlier compo- nents. This late negative wave is always reduced or absent during the conditioning period in experiments on OSPs at the low rates of 0.3-1 Hz. We have not enough cases to establish whether consistently at the same long ISI it is present when the subject has no task and absent when the subject is instructed to pay atten- tion to the ISis or to count the missing stimuli.

The OSP after low frequency conditioning flashes is a large, slow, positive wave, with a peak latency usually 500-1100 msec after the due-time of the missing flash or first missing flash after the end of a train (Fig. 4). Although this range in peak latency is large among subjects (mean 1002 + 143 msec for a set of experi- ments with ISI 1.7-3.2 sec), the latency is quite con- stant for a given subject at different ISis (S.D. < 100 msec is less than the uncertainty of measurement of these very slowly rounded maxima), as though it begins accurately at the moment a flash is expected and is then suppressed if a flash occurs. The beginning of the slow rise in positivity may have a very short latency after the due-time or even precede it, not as part of the SSR but unique to the interval during which the target is anticipated, since this is not a paradigm with random missing flashes.

We have not observed alternation in averaged VEPs (SSR) or OSPs in the low frequency range.

Effects of jitter of ISI on OSP Since the OSP is traditionally regarded as a sign of

expectation and behaves as though the system accu- rately detects the absence of a stimulus at the due-time, an obvious test is to manipulate the expectation by making the prediction of the next stimulus uncertain. In previously reported experiments with fish (Bullock et al. 1990) we found that a high percentage jitter of the ISis around the same (6 Hz) mean as a regular train that caused a good OSP, does not reduce the OSP substantially, as long as the last interval is the same as the mean interval. In human subjects, where only the low frequency range has been used heretofore, the effect of ISI jitter has been briefly mentioned by Takasaka (1985) who found it abolished the expecta- tion ERP. The instruction given to subjects in the present experiments was to try to predict the time of the next flash; we chose to use the end-of-train method since single omissions would be confounded by the

HUMAN VEP AND OSP 47

next VEP, therefore could not instruct subjects to count omissions.

At a mean conditioning stimulus frequency at the bottom of the high range (6 Hz), the OSP does not disappear totally with a rather high jitter, providing the train always ends with a mean ISI; this is also true at the high end of the range, e.g., 20 Hz (Fig. 5A). During the irregular train the SSR is quite irregular, large and small VEPs following long and short intervals. We did not study in human subjects, as we have done in fish (KaramiJrsel and Bullock 1994), the relative impor- tance of the last interval and of preceding intervals that can define a time constant of forgetting. This is be- cause it would require still more averaged scalp record- ings in addition to the already protracted time each subject is asked to sit.

At low mean frequencies (0.5 Hz), the OSP is wiped out or greatly reduced by jitter (Fig. 5B).

Effects of room light, flash intensity and stimulus type on VEP and OSP

OSPs are not greatly affected by flash intensity, but steady state VEPs become larger with high intensities; therefore the relative amplitude of OSPs gets smaller. Especially at the low end of the high frequency range (5 and 6.7 Hz) the OSP looks like a part of a large VEP. Ambient room light slightly reduces the VEP but hardly influences the OSP. For these reasons low flash intensities in moderate room light were used in this study.

P z

4

6.7 ~ ' ~ " ~ - - , r ~

O, r l M~ A AA AA ~,/q,, j - . . . . . .

I . . . . . . .

6.7 ~ ~ ' ~ ' ~ " ~ " " " ~ ' ~ ' ~ " I

l

I

. . . . ' . . . . 2 : ° c - - - • . . . .

- 3 J S e c o n d s S e c G n d s

Fig. 3. OSPs as function of ISI; simultaneous recordings from 4 sites: midline frontal, central, parietal and occipital. Traces are aligned by the last flash. Due-time of first omitted stimulus is shown by vertical dashed lines. Note that the best OSP is in Pz, mostly N170, P230 after the first missing stimulus due-time at high frequencies (10, 15 and 20 Hz). The P230 alone appears first at 5 Hz (Pz). The early components of the OSP responses are smaller in Oz whereas the SSR waves are larger. The late components of the OSPs are much the same in the 4 loci. Averages of 75

sweeps.

48 T.H. B U L L O C K ET AL.

Above 3 Hz there is another illusion: the flash with its diffuser held close to the face (10-15 cm), and without near accommodation, causes an impression that the regular flashing frequency is irregular. With strong attention this illusion can be kept for a long period (ca. 1 min). Under these conditions alternation of VEP amplitudes (every other VEP is larger) can be seen in the average (because the phase of the first alternation and the 1 in 2 ratio are consistent).

A 2nd Stim, /~ m ~X~ Next d u ~ j ~ f v ~ L . 3:12 :

I

_3 ~

1.7

~.2 2.4 3.5 4a 6.0 Seconds

B End of train~ ~, Next due F =

4 Hz

I

~ / ' V 6.7

L P 2 ~

1 25 2.5 Seconds

Fig. 4. OSPs as function of ISI, in the two ranges. A: long ISI range, 1.7-3.2 sec, regular flashes, omitting every third, i.e., conditioning with a single interval. Derivation from locus P3. Each trace the average of 100 sweeps, aligned by the time of the second flash, the one before an omission. Arrows show the due-t ime of the missing stimulus. Compare the sharp negative-positive VEP to the second st imulus with the broad, slow, positive OSP at 800-1100 msec after the due-time. B: high frequency range, 4 -20 Hz trains. Derivation from Pz; averages of 75 sweeps. Note the VEPs at 4 Hz alternate between large and small. There are only VEPs and no clear OSP < 5 Hz and only the P230 (measured from the due-t ime of the first omitted st imulus) at 5 and 6.7 Hz; a large N170 comes in at 10-20

Hz. Bandpassed 0.1-50 Hz.

A

Jittered ISI, mean = 0.75 s

Regular ISI = 0 . 0 5 s , ~ / X 3

- 3

B Jit tered ISI, m e a n = 2

ISI = 2 s Re.qular

° . . . . o l o . . . . . . . . . . . . 1 Due-time Sec onds

Fig. 5. Effect of jitter of ISis. A: in the fast st imulation range. Below: unjittered, regular train at 20 Hz. Above: mean ISI is 50 msec; jittered between 25 and 75 msec pseudorandomly with the condition that the last ISI is 50 msec. Two averages of 38 sweeps, superim- posed; each sweep is an identical t ime series of ISis; derivation Pz. The OSP is only slightly reduced by this jitter. Note the irregular responses to the irregular stimulation during conditioning. B: effect of jitter in the low frequency range; different subject and locus. Zero time is the due-t ime of the omitted stimuli; no stimulus markers are shown because no stimuli occur in the period illustrated. Below: a regular series of constant ISI = 2 sec, with every fourth to seventh flash omitted, causes the OSP shown, a large positive wave centered about P650. Above: when ISI is jittered around the mean = 2 sec, between 1.5 and 3 sec, pseudorandomly but always ending with 2 sec, the OSP is not convincingly present at all. Averages of 26; amplifier

filters 0.3-30 Hz; derivation P3.

Actually, it is not important to use the diffused flash. We find quite similar OSPs with a distant flash unit (ca. 2 m), subtending only a few degrees of visual angle. It is not necessary to use the flash unit, with its 10 cm diameter and relatively bright white light. Quite similar, small OSPs and SSRs to < 2 Hz can be ob-

H U M A N VEP A N D OSP 49

tained with small (3 mm) and weak light sources such as the green light emitting diode in the panel of the pulse generator that monitors the output triggers (modules of Pulsemaster A 300).

Effect of the type of attention on OSP

In order to show good fast OSPs (> 5 Hz), it is necessary to pay attention to looking at the same point on the diffusing screen of the flash unit; fixation of eyes is important. It is not necessary to count omissions or pay attention to the frequency; in any case subjects report it is difficult to follow > 3 Hz, consciously. Slow OSPs, in contrast, need some kind of conscious atten- tion, usually assured by instructing the subject to count the omissions. OSPs to low frequencies ( < 1 Hz) need only attention to the frequency, n o t to fixation at exactly the same point. It is not even necessary to look at the flash unit; the gaze can wander so long as the subject can see enough to follow the frequency. (See paragraph above for the light requirements.) The sec- tion Low stimulus rate ( < 2 Hz) described a marked effect of the instruction "Count the omissions" upon the standard VEP, reducing or abolishing the normal, large N250-350 of the VEP without attention.

Binocular interaction

Closing one eye or switching from one eye to the other during conditioning has no effect on the OSP at low frequencies, whereas under some conditions it can alter the SSR. Eye blink artifacts can be seen in an average without distortion of the OSP. In direct recordings from the optic rectum of rays (KaramiJrsel and Bullock 1994) we reported an interaction on switching from monocular stimulation of one eye to the other eye, affecting only late components of the low frequency VEP; we have not seen this in humans perhaps because it is difficult to get late components reliably from scalp recordings. We could not do inter- action experiments reproducibly at high flash rates.

Cross-modal property of slow OSPs

Slow OSPs are independent of changing the stimu- lus modality during the train (cross-modal condition- ing). With a specific paradigm instruction such as: "Do not pay attention to the type of stimulus, pay attention only to the frequency and try to catch the omitted ones," the same OSP can be obtained cross-modally or unimodally. At the end of the flash train, a click was added with the same ISI after the last flash, i.e., 2 sec. The same OSP wave sequence was obtained after the next due-time (Fig. 6), as though it were a flash, if we look at the scalp locus with the best OSP. A flash after a click train generates the same result. Fast OSPs,

AEP

C, v~P ~ osP

i i Click j

F l a s h t r a i n e n d i n g in c l i c k t Due-time

I 1 1 t F l a s h t r a i n o n l y

I

t I Click F l a s h t r a i n e n d i n g in c l i c k J

F l a s h t a i n i , p v 0 r o n - , r , , , , - - , ,

j ; 4 B - 5 S e c o n d s

Fig. 6. Cross-modal OSP. The last st imulus in a regular train of 7 flashes at 0.5 Hz is changed to a click in the upper traces for each locus, Cz and Oz, but not in the lower control trace. In both loci the click elicits its own auditory evoked potential (AEP) and in locus Cz but not in Oz the click acts just like a flash in preventing the OSP. After the due-t ime for the next, missing stimulus, the OSP appears

on schedule in Cz. Averages of 75 sweeps.

however, are not cross-modal; a click after a flash train or a flash after click train are new stimuli that cause EPs superimposed on the OSP which comes at the time expected if the click were absent.

I/EP and OSP in different scalp loci

The best scalp locus for the VEP is not always the best place for the OSP. The VEP and SSR are largest usually in occipital and sometime frontal derivations. The early components of the fast OSP are better in occipital while the late components are better in pari- etal regions and especially at the vertex (Fig. 3). Slow OSPs are generally large and clear at the vertex and at parietal loci.

Evidence for fluctuations in sensitivity

We do not usually see evidence for substantial fluc- tuations over time or recording sessions in respect to

50 T.H. B U L L O C K ET AL.

the measures studied. Similar VEP, SSR and OSP responses can be obtained from the same subject in different sessions on the same day or after several days or weeks if h e / s h e can maintain attention. Some sub- jects, however, showed no clear OSP at low frequencies in early recording sessions but did in later sessions. Full attention is absolutely needed for a good slow OSP. Such attention is not necessary for the fast OSP, if fixation is maintained.

Induced rhythms

In 4 subjects, as a late part of the OSP, an oscilla- tory component of 10-13 Hz was observed in the averages, at 10 and 15 Hz flash rates, lasting ca. 300 msec. It should be noted that in more favorable record- ing conditions, for example in the turtle (Prechtl and Bullock 1994), longer lasting induced rhythms, possibly comparable to the gamma band in mammals, consider- ing the body temperature, were seen in single sweeps but disappeared in averages due to frequency or phase jitter. It may well be that humans have such induced rhythms as a late component of the OSP, not perfectly time locked to the due-time of the omitted stimulus, and poorly seen in scalp recordings.

Discussion

A primary finding of this research and its cognates in non-mammalian subjects is that responses in the

brain to repetitive stimuli cannot be understood simply in terms of refractoriness and recovery. Processes des- ignated - by their outcome - as facilitation, suppres- sion, rebound and induced rhythms are some of the most obvious additional elements. They may have rates of onset and decay not only different from each other but distinct for the earlier and later components of the response and for responses to faster and slower condi- tioning trains.

As others have noted over the years, endogenous and exogenous events in the brain form a spectrum with intermediates. We would underline that the neu- robiology of expectation - of which the omitted stimu- lus paradigm is a special case - covers a variety of processes, probably a variety of mechanisms, at lower and higher levels in the brain, some even quite periph- eral (Bullock 1988, 1993). Since the search for an omitted stimulus ERP in lower vertebrates revealed one already in the retina (Bullock et al. 1990; Prechtl and Bullock 1994) and in medullary nuclei for a non- visual modality (Bullock et al. 1993), but only in the "high" frequency range (> 2 Hz), heretofore unknown in experiments on humans, it became interesting to test for such an ERP in humans.

Another primary finding is that two kinds of OSPs are emitted after conditioning in different ranges of flash rates (Table I). Not only do we find a missing stimulus ERP in the high frequency range in the pres- ent experiments, but we find this range is not continu- ous with or even overlapping with the conditioning frequency range used by previous authors for the famil-

T A B L E I

Comparison of the two kinds of visual OSP in humans . OSPs follow the end of flash trains or single omissions in the frequency ranges shown; note the gap between the frequency ranges where no OSP is found. "Fas t" and "slow" refer to the rates of conditioning flashes.

"Fas t" OSP "Slow" OSP

> 2 Hz (human data: 5 - 4 0 + Hz) < 2 Hz (human data: 0.3-1.6 Hz) Frequency range of conditioning flash series (large, diffuse or small, weak source)

Latency of first main peak after due-t ime a; polarity and msec

Required state Multimodal: click is equivalent to flash Probable origin Effect of jitter of IS Ie Required conditioning series

Best scalp locus Binocular interaction

N120, P170-230, plus tendency to a late (300 msec) induced rhythm (10-13 Hz) Fixation but not attention c No Ret ina d Slight reduction of OSP 250-350 msec or < 5 flashes; repeated for averaging f Occipital Does not exist

P500-1100 b

Attent ion but not fixation Yes Unknown Severe reduction of OSP 2 flashes; repeated for averaging g

Vertex and parietal Does exist

a Latency tends to be the same for any conditioning frequency within the range. b There may not be a difference in the tendency to a late induced rhythm since slight jitter of its latency, phase or frequency would reduce it greatly in averages. c Fixation of gaze upon the light source. Attent ion to the timing of the omission. See instructions to subjects. d Based on experiments on rays and turtles (Bullock et al. 1990; Prechtl and Bullock 1994). e See text for details of the mean frequency and range of jitter. f This is not necessarily different from the slow OSP. Due to the shorter latency, 5 is about the min imum to satisfy the definition of an OSP - that it cannot be considered part of the last VEP. g The OSP appears in averages 500-1100 msec after the due-t ime of the missing third flash.

HUMAN VEP AND OSP 51

iar human ERP to missing stimuli, which is a low frequency range, < 2 Hz. We have no good suggestion for an explanation of this blind spot in the spectrum but regard it as more likely apparent than real. The absence of OSPs in a range of conditioning flash rates between the "high" (> 2 Hz) and the "low" ( < 2 Hz), actually between ca. 2 and 5 Hz, is curious and invites, firstly, renewed effort to find the conditions that will reliably elicit expectation potentials - since the other two ranges show such large differences in the condi- tions conducive to their OSPs. Of course, absence could mean simply that there is too much latency fluctuation to survive averaging. Alain et al. (1989) used 1.7 Hz tone stimuli, at the boundary between the "low" and the absent ranges; they reported very small or unclear OSPs in nearly half their subjects; in the better 4 out of 7 subjects only some electrode loci (temporal lobe) gave OSPs, smaller on average than ERPs to targets. Their OSPs were quite unlike ours in the same range, namely P350-400 instead of the much later maximum (P500-1100) in the present report. We have not done enough with auditory OSPs to be able to say whether this difference in modality could explain the difference in latency.

In a number of dynamic properties, the two forms of OSP to two distinct frequency ranges of conditioning stimuli are quite different, including dependence on attention and on fixation, jitter of the conditioning intervals and the acceptance of stimuli of other modali- ties as part of the regular, expected series that prevents the OSP after each stimulus. Induced rhythms are a labile, late component of OSPs, especially of the fast type, in each species studied, sampling fish and reptiles as well as humans. These contrasts suggest that at least some parts of the mechanisms are distinct. Humans have both types; the non-mammals examined among fish and reptiles have so far revealed only the high frequency range OSPs. It is now of interest to learn whether the only obstacle to demonstrating the slow type in non-mammals is building in some attention, for example by reinforcing the omissions with classical conditioning.

Characteristic of both types is the surprisingly small number of conditioning intervals necessary to establish the temporally specific apparent expectation. Even af- ter a single interval (2 stimuli; Fig. 4A) a wave occurs at the characteristic latency for unequivocal OSPs that satisfy the defining condition: being not attributable to VEPs.

In our fish and reptiles we have found the fast OSP is already present in the retina and repeated, with modification in the midbrain, from which one inter- prets this ERP to be not a high level cognitive expecta- tion. We have called it evidence of an "apparent ex- pectation," because of the tendency to a constant la- tency over a wide range of ISis (Bullock 1988, 1993),

and offered an explanation based on low level proper- ties that could even belong to receptors (Bullock et al. 1990, 1993; Prechtl and Bullock 1994). What about the slow OSP; at what level does it appear? As part of the assessment of its cognitive level, it will be important to find the means to discover whether the slow OSP begins in the brain-stem or even the retina in humans. It would also be desirable to know whether the fast OSP is similar to that of the sample species of fish and reptiles in arising in the retina.

Parenthetically, it should be pointed out that we repeatedly refer to the retina only because the present study was based on visual stimuli. "Earliest afferent stations and possibly even receptors" would be more informative, since we found OSPs in another modality, electroreception, in elasmobranchs, in the first medullary nucleus and some signs of them in the primary afferent axons (Bullock et al. 1993).

VEPs are noteworthy, even without dwelling on the marked diversity among human subjects in the ampli- tude and form, which are already familiar in the litera- ture, or upon the transition from the rested or long ISI form to that of the SSR, a transition which is not yet adequately described, let alone understood. Alterna- tion of successive VEP amplitudes (large, small, large, small . . . = " 1 in 2") at certain flash rates (> ca. 4 Hz), aided by conditions of attention and looking, is a striking dynamic property that we have been able to explore only in a preliminary way. We are reminded of the quite consistent alternation in elasmobranchs (KaramiJrsel and Bullock 1994), extending from 1 in 2 up to a large VEP every fourth and occasionally every seventh flash. Such high ratios as 1 in 7 belong in the same domain as the evoked responses to flashing be- tween the uppermost following frequency (UFF; 1:1 without averaging) and the lowermost fusion frequency (LFF; no ripple even in the average). While numerical values for these landmarks depend on many factors, the point of interest revealed by evoked potentials is that, in this frequency zone, there is no steady state but an irregular succession of small EPs or, on occasion a regular, high ratio. We report here that alternation is not unknown in humans in the same frequency range, even in averages. A few observations are recorded on illusions seen by subjects during alternating VEPs.

Our interpretation of the fast OSP, which in our fish and reptiles begins already in the retina, is quite differ- ent from the usual view of the slow OSP. As we have proposed elsewhere (Bullock et al. 1990; Karamiirsel and Bullock 1994), this "apparent expectation" might mean that the equilibrium of inhibition and excitation achieved during the conditioning train automatically makes the latency of onset of the excitatory process just barely longer than the last ISI so that if the inhibition contributed by a stimulus is missing or de- layed by as little as 5 -7 msec, an after-discharge pro-

52 T.H. BULLOCK ET AL.

cess equivalent to a stimulus proceeds unhindered to consume its characteristic latent period for firing the OSP.

The SSR and the OSP represent two kinds of short term memory. We conclude that a major category of sensory response characteristics, besides the classical ones (sensitivity, receptive fields, recovery times and the like), is that of dependence upon recent history of iterative events, including their intervals, delays and omissions; the dependences we speculate are multiple and partly independent, for example multiple facilitat- ing and forgetting time constants. Such a conclusion means that new effort is needed to study single sweeps (Achimowicz 1991) and the relation of responses to the prestimulus or pre-event background activity. The present study has examined parametrically only some variables of recent history. Such dynamic characteris- tics are important neglected properties of afferent sys- tems at each level from the receptor onwards.

Some of these dynamic properties might be found useful in characterizing the state of the brain in patho- logical conditions, in recovery of function after insults, and in stages of development and aging. Presumably there will be a trade off between robustness and sensi- tivity that will be different, for example, as between fast and slow OSPs. The reason for so thinking is that these properties are probably dependent on non-linear interactions of two or more time varying processes, sensitive to state and history as well as input. It re- mains to be learned whether the most useful combina- tion of robustness and sensitivity for a particular condi- tion will be the OSP latency or amplitude, lowest conditioning frequency, amplitude increase with condi- tioning frequency, induced rhythm, UFF, LFF or some derivative of these.

T.H.B. was aided by grants from the National Institute of Neuro- logical Diseases and Stroke, the National Science Foundation and the US Air Force Office of Scientific Research. S.K. was supported by the Department of Physiology, Istanbul Medical Faculty, and Electro-Neuro-Physiology Research and Application Center, Univer- sity of Istanbul, Turkey. C.B.-E. was supported by the German Science Foundation.

Some of the experiments were done in the laboratory of Prof. E. Ba§ar.

References

Achimowicz, J.Z. On the deterministic brain states revealed by VEP classification in phase domain. In: I. Dvorak and A.V. Holden (Eds.), Mathematical Approaches to Brain Functioning Diagnos- tics. Manchester University Press, Manchester, 1991: 209-230.

Alain, C., Richer, F., Achim, A. and Saint Hilaire, J.-M. Human intracerebral potentials associated with target, novel, and omitted auditory stimuli. Brain Topogr., 1989, 1: 237-245.

Barlow, J.S., Morrell, L. and Morrell, F. On evoked responses in relation to temporal conditioning to paired stimuli in man. MIT Res. Lab. Electron. Quart. Prog. Rep., 1965, 78: 263-272.

Bullock, T.H. The comparative neurology of expectation: stimulus

acquisition and neurobiology of anticipated and unanticipated input. In: J. Atema, R.R. Fay, A.N. Popper and W.N. Tavolga (Eds.), Sensory Biology of Aquatic Animals. Springer, New York, 1988: 269-284.

Bullock, T.H. How Do Brains Work? Papers of a Neurophysiologist. Birkh~iuser, Boston, MA, 1993: 204-215.

Bullock, T.H., Hofmann, M.H., Nahm, F.K., New, J.G. and Prechtl, J.C. Event-related potentials in the retina and optic rectum of fish. J. Neurophysiol., 1990, 64: 903-914.

Bullock, T.H., Karamiirsel, S. and Hofmann, M.H. Interval-specific event related potentials to omitted stimuli in the electrosensory pathway in elasmobranchs: an elementary form of expectation. J. Comp. Physiol. A, 1993, 172: 501-510.

Courchesne, E. From infancy to adulthood: the neurophysiological correlates of cognition. In: J.E. Desmedt (Ed.), Cognitive Compo- nents in Cerebral Event-Related Potentials and Selective Atten- tion. Karger, Basel, 1979: 224-242.

Ford, J.M. and Hillyard, S.A. Event-related potentials (ERPs) to interruptions of a steady rhythm. Psychophysiology, 1981, 18: 322-330.

Hillyard, S.A. and Picton, T. Electrophysiology of cognition. In: F. Plum (Ed.), Handbook of Physiology: the Nervous System. Vol. V, Part 2. Higher Functions of the Brain. American Physiological Society, Bethesda, MD, 1987: 519-584.

Karamiirsel, S. and Bullock, T.H. Dynamics of event-related poten- tials to trains of light and dark flashes: responses to missing and extra stimuli in rays. Electroenceph. clin. Neurophysiol., 1994, 90: 000-000.

Kerkhof, G.A. Decision latency: the P3 component in auditory signal detection. Neurosci. Lett., 1978, 8: 289-294.

Klinke, R., Fruhstorfer, H. and Finkenzeller, P. Evoked responses as a function of external and stored information. Electroenceph. clin. Neurophysiol., 1968, 25: 119-122.

Mangun, G.R., Hillyard, S.A. and Luck, S.J. Electrocortical sub- strates of visual selective attention: synergies in experimental psychology, artificial intelligence, and cognitive neuroscience. In: D.E. Meyer and S. Kornblum (Eds.), Attention and Performance XIV. MIT Press, Cambridge, MA, 1993: 219-243.

McCallum, W.C. Brain slow potential changes elicited by missing stimuli and by externally paced voluntary responses. Biol. Psy- chol., 1980, 11: 7-19.

Picton, T.W., Hillyard, S.A. and Galambos, R. Cortical evoked responses to omitted stimuli. In: M.N. Livanov (Ed.), Major Problems of Brain Electrophysiology. Academy of Science, Moscow, 1974: 302-311.

Prechtl, J.C. and Bullock, T.H. Barbiturate sensitive components of visual ERPs in a reptile. Neuroreport, 1992, 3: 801-804.

Prechtl, J.C. and Bullock, T.H. Event-related potentials to omitted visual stimuli in a reptile. Electroenceph. clin. Neurophysiol., 1994, in press.

Renault, B. and Les~vre, N. Topographical study of the emitted potential obtained after the omission of an expected visual stimu- lus. In: D.A. Otto (Ed.), Multidisciplinary Perspectives in Event- Related Brain Potential Research. National Technical Informa- tion Service, Springfield, VA, 1977: 202-208.

Renault, B. and Les~vre, N. A trial by trial study of the visual omission response in reaction time situations. In: D. Lehmann and E. Callaway (Eds.), Human Evoked Potentials. Plenum Press, New York, 1978: 317-329.

Ritter, W., Simson, R. and Vaughan, H.G. Topographic analysis of task-related cerebral potentials. In: J.E. Desmedt (Ed.), Cognitive Components in Cerebral Event-Related Potentials and Selective Attention. Karger, Basel, 1979: 132-139.

Ruchkin, D.S. and Sutton, S. Latency characteristics and trial-by-trial variations of emitted cerebral potentials. In: J.E. Desmedt (Ed.), Cognitive Components in Cerebral Event-Related Potentials and Selective Attention. Karger, Basel, 1979a: 106-118.

HUMAN VEP AND OSP 53

Ruchkin, D.S. and Sutton, S. CNV and P300 relationships for emitted and for evoked cerebral potentials. In: J.E. Desmedt (Ed.), Cognitive Components in Cerebral Event-Related Poten- tials and Selective Attention. Karger, Basel, 1979b: 119-131.

Rusinov, V.S. General and localized alterations in the electroen- cephalogram during the formation of conditioned reflexes in man. Electroenceph. clin. Neurophysiol., 1960, 13: 309-319.

Squires, N,K., Squires, K.C. and Hillyard, S.A. Two varieties of long latency positive waves evoked by unpredictable auditory stimuli in man. Electroenceph. clin. Neurophysiol., 1975, 38: 387-401.

Squires, N.K., Squires, K.C. and Hillyard, S.A. Functional equiva-

lence of signal-present, signal-absent, and threshold-detect P3s. In: D.A. Otto (Ed.), Multidisciplinary Perspectives in Event-Re- lated Brain Potential Research. National Technical Information Service, Springfield, VA, 1977: 218-221.

Sutton, S., Braren, M., Zubin, J. and John, E.R. Evoked-potential correlates of stimulus uncertainty. Science, 1965, 150: 1187-1188.

Sutton, S., Tueting, P., Zubin, J. and John, E.R. Information delivery and the sensory evoked potential. Science, 1967, 155: 1436-1439.

Takasaka, Y. Expectancy-related cerebral potentials associated with voluntary time estimation and omitted stimulus. Folia Psychiat. Neurol. Jpn., 1985, 39: 167-172.