Late components of saccade-related brain potentials in guessing tasks

652

Electroencephalograptzv and clinical NeurophysioloKv , 1983,56:652-663

Elsevier ScientificPublishers Ireland, Ltd.

LATE COMPONENTS

O F S A C C A D E - R E L A T E D B R A I N P O T E N T I A L S IN G U E S S I N G T A S K S

MAGDA MARTON, JOZSEF SZIRTES and PETER BREUER Institute for P.~vchologv, Hungarian Academy of Sciences, 1394 Budapest 62 (Hungao')

(Accepted for publication: June 9, 1983) In describing the way human organisms interact with their environment modern cognitive psychology has put strong emphasis on the active role of the perceiver in continuously constructing hypotheses about future events around him (Neisser 1976). When the expectations are violated the surprising stimuli lead to a revision of the perceiver's hypotheses. It was proposed (Donchin 1975, 1979) that this kind of hypothesis updating is reflected in the late positive component (P3 or P300) of the human event-related potentials (see also Ritter 1978). This suggestion also receives support from studies describing the relation between subjective probability, task information and 'equivocation' on the one hand, and the amplitude of the P300 component on the other (for a review see Johnson 1983). At present, it seems to be an open question whether or not comparable late 'endogenous' positive components could be observed in experiments in which the subject must actively search for taskrelevant visual information. In normal conditions the environment is regularly scanned by saccadic eye movements and, because of the suppression of pattern vision during saccades, visual information is picked up and processed only during the fixational pauses between saccades. Therefore it seemed possible that cortical responses time-locked to saccades (i.e., lambda responses) might offer an important additional measure, besides evoked potentials, to study visual information processing. Systematic experiments with human subjects demonstrated that the lambda response accompa-

Address correspondence to: Magda Marton, Ph.D., Institute for Psychology,HungarianAcademyof Sciences,H-1394 Budapest 62, Pf. 398, Hungary.

nying saccades consists of a sequence of components some of which are associated with the onset, others with the offset of the saccade (Kurtzberg and Vaughan 1977; Yagi 1979). We obtained comparable results when using rhesus monkeys as subjects (Szirtes et al. 1982). It is now widely accepted that components appearing after saccade offset, especially the positive peak at around 100 msec, can be viewed as an evoked potential elicited by the stimuli appearing at the new fixation (Gaarder 1975; Kurtzberg and Vaughan 1977; Yagi, 1979, 1981 ; Scott et al. 1981). There are only a few studies investigating taskdependent modifications of the late lambda components. Cooper et al. (1977, 1978) reported that detection of an infrequent target in a complex visual display was accompanied by a large positive wave with an amplitude maximum at vertex and parietal areas. A saccade leading to the target area was found to precede this 'detection positivity' by 200-300 msec. The authors suggested that this positivity reflects similar processes to the P300 component of the visual event-related potentials (ERPs). In an earlier study we demonstrated that, when a small pattern is switched on at the moment of the eye's arrival at its new fixation position, there is an increase in the amplitude of a component appearing at around 340 msec after saccade onset (Marton et al. 1983). Kurtzberg and Vaughan (1979) reported that, in comparison to pattern inspection, the magnitude and latency of the main positive lambda components change, when responses are recorded during verbal or spatial tasks. Yagi (1982) compared lambda responses appearing either during inspection or during searching of a map. Four of the 6 subjects studied showed a larger amplitude of the positive-negative components at 100-180 msec after saccade offset in the

0013-4649/83/$03.00 ¢~ 1983 ElsevierScientificPublishers Ireland, Ltd.

LAMBDA RESPONSE AND GUESSING search condition than during simple inspection of the map. Further experimental observations are needed, however, to answer the question whether the late components of the lambda response also reflect the effect or consequences of information processing, especially in cognitive tasks where this processing is closely linked to saccades. For an effective investigation of this question it would be useful to modify one of the paradigms used for studying late ERP components in such a way that perceiving task-relevant information would require saccadic eye movements. The large-amplitude, positive parieto-central component of ERPs with a latency of 300-500 msec (usually called P300 or P3b) occurring in response to infrequent a n d / o r task-relevant stimuli was discovered by Sutton et al. (1965, 1978). Since then their results have been replicated in different experimental contexts and in different laboratories (Donchin and Cohen 1967; Ritter and Vaughan 1969; Hillyard et al. 1971; Courchesne et al. 1975; for reviews of the literature on P300 see Hillyard and Picton 1978; Pritchard 1981). The late positive component of ERP is almost always associated with a 'slow wave' (SW) component (N.K. Squires et al. 1975, 1977; K.C. Squires et al. 1977). Principal component analysis revealed that the slow wave begins at around 300 msec and may last for more than 1 sec. It is positive in the posterior scalp regions but negative at frontal electrodes. The P300 and the slow positive parietal wave have been observed to follow both infrequent or task-relevant stimuli, although during discrimination their dissociation might occur (Roth et al. 1978; Ruchkin et al. 1980; Parasuraman et al. 1982). For the present experiment we looked for a paradigm which would be certain to elicit these late positive components. It seemed that the paradigm in which subjects have to guess what the ensuing stimulus would be might serve just this purpose (Tueting et al. 1970; Friedman et al. 1973; Ruchkin et al. 1975). When in such studies stimulus probability was manipulated the P300 amplitude varied inversely with the joint probability of the stimulus and guess (outcome probability). A common element of such experiments is that the

653 stimulus provides information which resolves the subject's uncertainty inherent in his guess. Furthermore, the feedback stimulus also modifies the subject's expectations concerning future trials, and these changes are also reflected in the P300 (Sutton et al. 1978). In our present experiment we used a guessing paradigm modified in such a way that perception of task-relevant information required a saccadic eye movement. This condition seemed to provide an opportunity to explore the way the late lambda components reflect information processing. Specifically, we wanted to describe the late components accompanying the cognitive evaluation of information picked up by saccadic eye movements. Furthermore we wanted to see whether principal component analysis reveals similar underlying basic wave forms as found for the late components of traditional ERPs (that is, ERP recording when the eye is held stationary).

Method

Subjects and procedure Nine subjects (3 female, 6 male, mean age 27 years) with normal or corrected-to-normal vision were paid to take part in the experiment. Six of the subjects had participated in other ERP experiments before, but none was aware of the purposes of the present study. Each subject was seated in a sound-attenuating chamber facing a screen placed 1 m in front of him. Head movement of the subject was restrained by a headrest. The only source of light in the room was a continuously illuminated subminiature fixation lamp (0.2 ° ) located at one side of the screen 22 ° away from the midline. There were two conditions: control and guessing. In both conditions the trials began by the subject fixating the continuously illuminated fixation lamp. In the control condition the subject's task was first to perform a saccade to a stimulus light at the middle of the screen when it was switched on. Then, after 2 sec, the target stimulus (i) was switched on and at that point the subject had to make a further saccade in the same direction to this target area located 22 ° from the midline. At the target area there were 3

654

subminiature lamps in a row, separated by a visual angle of 1.6 ° from one another. Subjects were presented with 6 blocks of 25 trials. In three of these they had to perform saccades from left to right, in the other three in the opposite direction. Within a given block of trials the number of lights at the target location remained unchanged, that is, for a given block subjects always knew beforehand whether 1, 2 or 3 lights would appear at the target area. In the guessing condition the subject's task was first to make a verbal guess about how many target lights would appear on a given trial while fixating the continuously illuminated fixation lamp. After guessing subjects made a saccade to the middle lamp. Then, after 2 sec, when the target stimulus was switched on subjects had to make a further saccade to the target. The target stimuli appeared with unequal probabilities in a computer simulated randomized sequence. Each sequence consisted of 84 trials. The different stimulus probabilities were 17, 33 and 50%. The three kinds of target were assigned different probabilities and the respective probabilities were regularly permutated from sequence to sequence. At the end of the sequence the subject had to judge which was the most and which was the least frequent target stimulus. No reward or penalty was given for correct or incorrect guesses. At half-time, in both conditions the position of the fixation point and the target stimuli was reversed. Thus the subjects performed equal numbers of saccades in both horizontal directions.

M. MARTON ET AL.

lambda responses. Records were edited before averaging, i.e., trials with blinks, movement artifacts or poorly performed saccades were left out from the analysis. Equal numbers of lambda responses associated with left and right saccades, respectively, were included in the individual average wave forms. The analysis epoch for lambda responses extended from 172 msec prior to saccade onset to 852 msec after it. The sampling rate was 4 msec per point. In the case of the contingent negative variation (CNV) averages, when both the first and second saccade and the interval between them were included, the analysis epoch extended from 172 msec prior to the first saccade to 2900 msee after it. The sampling rate was 12 msec per data point. Average wave forms of 50-80 sweeps were written out on an X-Y plotter. Component peak latencies were measured relative to saccade onset, while peak amplitudes were relative to the baseline, defined as the average amplitude of EEG activity preceding saccade onset. Average wave forms were also subjected to Principal Component-Varimax Analysis (PCVA) to deal with the possible overlaps of underlying components. The 108 averages were the input to the PCVA (2 conditions × 6 electrodes × 9 subjects). The wave forms were condensed to 64 points by using every fourth data point only. For each individual average the mean EEG activity preceding saccade onset served as the baseline. Then a 64 × 64 time point covariance matrix was computed and factored using the SPSS program package (Nie et al. 1972) adapted for a TPA 1 lf140 computer.

Recording and data analysis Silver-silver chloride electrodes were affixed with collodion on midline scalp sites (Fz, Cz, Pz, Oz) and also at points P3 and O1 according to the 10-20 system. The electro-oculogram (EOG) was recorded between electrodes at the outer canthus of each eye and above and below one eye. Brain activity and the EOG were amplified and stored on FM tape. Unipolar records were made, linked ears serving as reference. The overall bandpass of the system was 0.16-1250 c/sec. Averaging was performed off-line on a small computer (TPAi, K F K I ) ; the onset of the second saccade leading to the target area served as the trigger for averaging

Results The mean durations of saccades to the target area were not significantly different between the control and guessing conditions (58.9 versus 56.4 msec). A two-way analysis of variance (ANOVA) was applied separately to the latency and to the amplitude for components which could be reliably measured after saccade onset in each case. The two factors were condition and electrode location. ~We note that in neither ANOVA was there a significant interaction between the condition and electrode factors.

LAMBDA RESPONSE A N D G U E S S I N G

655

The first peak, a 'spike-like' positivity (PO) was most probably a reflection of volume conducted eye muscle activity (Szirtes et al. 1982). Following the PO, a negative component appeared at about 45 msec (N45) from saccade onset (Fig. 1). An A N O V A for N45 amplitudes revealed a significant effect for condition ( F (1, 96) = 4.26, P < 0.05). The increase in magnitude of the N45 between the control and the guessing condition was most

Cz

Z' ¢~'~,,

.........

.....

"..

•

,-"

Pz •

...Y"

••

""..

z

"... P375

..."

%''-,

OZ" "'"""

• ....

N45 1

......... " ..........""

-.....N~;0

..

0

.....'"'"""

~

N220

'.- .%

E O G - ~

..j

176 ..................... P440

5•uV +

1OO msec --

CONTROL

............ GUESSING Fig. 1. Grand mean lambda response averaged across all subjects in the control and guessing conditions. The bottom curve illustrates the averaged EOG activity of saccades in one horizontal direction.

pronounced over O1, the lateral occipital area ( - 1.92 vs. -3.58/~V). The second negative peak appearing at around 110 msec after saccade onset ( N I l 0 ) could be observed primarily in the occipital records. This negative component showed a tendency to peak earlier and to be smaller in the guessing condition; however, these differences did not reach significance. The positive peak appearing at around 176 msec after saccade onset in the occipital leads is the classical ' l a m b d a wave.' The amplitude of the P175 was significantly greater in the guessing condition than in the control condition ( F ( 1 , 96)= 8.30, P < 0.01). Both the P176 and especially the subsequent N220 emerged most clearly in the occipital responses (see Fig. 1). It is possible that in the centro-parietal derivations the late positive components already overlap the P176 and N220 peaks. We postpone the discussion of this problem of overlap until we turn to the results of PCVA for the lambda responses. The most striking difference between the two conditions appeared in the late positive complex or component. In the control condition the peak latency of this positivity generally appeared at around 300 msec following saccade onset, while in the guessing condition it peaked about 60-120 msec later ( F ( 1 , 9 6 ) = 15.97, P < 0.001). The amplitude of the late positivity also showed a significant difference between the two conditions ( F ( 1 , 9 6 ) = 40.27, P < 0.001) and among the electrode positions ( F (5, 96) = 3.31, P < 0.025). The greatest amplitude appeared in both conditions over the parietal region (Table I). However, in the control condition the second largest peak was recorded occipitally, while during guessing it was found centrally.

Principal Component-Varimax Analysis Four factors were extracted by the PCVA of the lambda responses to the target stimuli, accounting for 79.6, 8.0, 4.3 and 2.4% of the variance of the deviation wave forms (i.e., the differences between each individual average and the centroid), respectively. The four factors were rotated, using the normalized varimax criterion. The loading pat-

656

M. MARTON ET AL.

TABLE 1 Means (,~) and standard deviations (S.D.) for the peak latency and the amplitude values of the late positive component in the control and guessing conditions. N = 9. Electrode Fz

Cz

Pz

P3

Oz

O1

X S.D.

291 99

297 102

314 79

319 72

388 46

322 57

Guessing X, S,D.

318 95

365 83

375 60

444 56

437 51

439 36

Latency (in reset)

Control

Amplitude (in llV)

,X S.D.

0.8 5.5

3.8 5.2

6.4 4.0

5.8 2.9

6.3 2.8

6.2 2.9

Guessing X S.D.

7.7 6.9

13.1 8.4

16.8 7.2

13.3 6.0

10.1 5.4

10.6 6.8

Control

FACTOR 1

FACTOR 2

FACTOR 3

FACTOR 4

I

-200

0

I

I

I

200

400

600

I

800 msec

Fig. 2. Factor loadings for the 4 factors extracted from the PCVA of the deviation wave forms, computed from the covariance matrix of lambda responses. The polarities are adjusted so that the major peak of each basic wave form corresponds to the polarity of the original data wave forms. The vertical line indicates saccade onset.

terns for these four factors are depicted in Fig. 2. The loading p a t t e r n (or basic wave form) is a measure of the strength of association between a principal c o m p o n e n t a n d the original data points. Factor scores were also derived to assess the magn i t u d e of the c o m p o n e n t s in different l a m b d a responses (Fig. 3), Separate A N O V A s were carried out for the scores of each factor. ~ Factor 1 was a b r o a d factor which had a welldefined peak loading with a latency at a r o u n d 380 msec after saccade onset. A highly significant condition m a i n effect emerged in the A N O V A for the scores of the first factor ( F ( I , 9 6 ) = 3 7 . 8 1 , P < 0.001). This resembled the difference found between the two c o n d i t i o n s in baseline-to-peak values of the late positivity, reported above. In the guessing c o n d i t i o n this factor was the largest at the parietal area a n d loaded positively everywhere except the frontal site. The m a i n effect of electrode position was also significant ( F (5, 96) = 4.78, P < 0.01). O n the basis of its latency a n d scalp topogr a p h y factor 1 can be assumed to reflect processes traditionally identified as the P300 c o m p o n e n t or P3b peak in ERPs. Factor 2 began at a r o u n d 180 msec after sac-

2 In neither case was there a significant interaction observed between condition and electrode location.

LAMBDA RESPONSE A N D G U E S S I N G

657

FACTOR 2

FACTOR 1 20

1.0

0

o

H ~ Cz

-1D-

-1.o ~

El---E] Oz

FACTOR 4 / ~ 2.0- ~ ,

FACTOR 3 1,0

01

-2.0 ~ -1. CONTROL GUESSING I

I

CONTROL GUESSING

Fig. 3. Across-subject averages of the factor scores, as a function of experimental conditions, obtained from the PCVA of lambda responses. The ordinate scale is in arbitrary units.

cade onset and showed increasing loading values in the later part of the analysis epoch. Its time course was very similar to the slow wave factor identified in event-related potentials by several authors (e.g., K.C. Squires et al. 1977; Friedman et al. 1980). The ANOVA for factor 2 scores showed the main effect for condition ( F ( 1 , 96)=21.93, P < 0.001) and electrode position ( F (5, 95)= 3.97, P < 0.01) to be significant. In the guessing condition the topography of factor 2 scores was similar to that of factor 1, the P300 factor. Factor 3 reached its peak at around 95 msec after saccade onset. Both the effects of condition ( F (1, 96) = 4.99, P < 0.05) and of electrode position ( F ( 5 , 96) = 3.14, P < 0.05) were significant.

The distribution of score values between electrode locations demonstrates that this factor contributed most to the occipital records, and that its effect was weakest over the central region. On the basis of this topography factor 3 can be related to the N l l 0 component of the lambda response. The behavior of factor 3 might help to resolve the question of whether the decrease of N I l 0 in the raw potential data at the central and parietal lambda responses was due to the early beginning of the late positive processes. Since the results of the PCVA reflect distinct sources of experimental variance, the topographical distribution of the scores of factor 3 suggests that the presence of the N110 component in the occipital, and its absence from the central, leads during guessing is not related to an early appearance of a prominent positive process. Factor 4 had a characteristic peak at 178 msec after saccade onset, and can be identified with the positive lambda wave appearing at around 176 msec in the average lambda response, as depicted in Fig. 1. The ANOVA for factor 4 scores showed a highly significant effect both for condition ( F ( 1 , 96)= 38.79, P < 0.001) and for electrode ( F ( 5 , 9 6 ) = 6 . 5 2 , P < 0 . 0 0 1 ) . According to the scores in the guessing condition this factor provides the greatest contribution to the parietal records. This contribution is obscured, however, by the temporal overlap of the more powerful P300 and slow wave components. This explains why no clearly discernible P176 appears in the individual and the grand mean parietal responses. The effect of possible differences in anticipation and preparation on lambda responses between the two conditions was examined by making the averages (CNVs) include both the first and second saccades and the time interval between them. As shown in Fig. 4, although there are slight differences in the magnitude of the CNVs between the two conditions, the appearance of the P300 component in response to the target stimuli during guessing does not seem to be explicable on the basis of these differences alone. The visual impression of independence between CNVs and the late positivity of the lambda responses was confirmed when a PCVA was applied to the CNV wave forms. The first three factors

658

2°2;2°

M. MARTON ET AL.

.............

FACIO ~

......

FACTOR2 ,

\/

FACTOR3

..... /

Om

ec !

0 Fig. 4. Grand mean CNV wave forms averaged across subjects in the control and guessing conditions. The bottom curve illustrates the averaged EOG activity of saccades in one horizontal direction.

accounted for 61.4, 13.6 and 8.2% of the variance (Fig. 5). Factor 1 can be identified as the CNV itself. Factor 2 corresponds basically to the P176 component, which, in turn, can be identified with the classical lambda wave. Finally factor 3 is identifiable as the late positive component, the P300. The distribution of factor scores, shown in Fig. 6, also reinforces this interpretation. Factor 1, the CNV factor, revealed the highest loading over the frontal area and the smallest over the occipital region. Its loading peaked at around 1360 msec after the first saccade and reached its minimal value at around 190 msec after the second saccade. Neither condition nor electrode effects reached significance in the ANOVAs for factor 1 scores. The fact that factor 1 showed smallest loading

g

, I

1000

o

I

2000

o

I

3000 msec

Fig. 5. Factor loadings for the 3 factors obtained from the covariance matrix of the CNV wave forms recorded in the two conditions. Onset of the first saccade to the middle light stimulus at 0 msec. The polarity conventions are the same as in Fig. 2.

values in the time period in which both the P176 and the P300 factors began to show significant loadings suggests that these components are independent from the process of CNV resolution. Factor 2 showed loadings in those time ranges of the analysis epoch in which normally the classical lambda waves (the P176 components measured from onset) would be expected to appear following the first and second saccades. The loading of factor 2 showed a first peak at around 184 msec after the first saccade, while its second peak appeared at around 200 msec after the second saccade. The difference in loading magnitude between these two peaks might be due to the fact that the averages were time-locked to the first saccade (as

L A M B D A RESPONSE A N D G U E S S I N G

FACTOR

1

659

FACTOR

2

1.0

0

0

I

I

following the second saccade. The peak latency of the loading appeared at about 360 msec from the onset of the second saccade. This factor clearly corresponds to the late positive component, the P300. While an A N O V A for factor 3 scores revealed no significant main effects, as Fig. 6 shows, factor 3 again provided the greatest contribution to the parietal records.

Discussion

-0.5-

-1.0-

I

I

CONTROL GUESSING

FACTO R 3 1.0"

H 0

Fz 0

Cz

-'P3 []

[] Oz 01

-1.0-

-2.0. I

I

CONTROL

GUESSING

Fig. 6. Factor scores averaged across subjects corresponding to the 3 factors depicted in Fig. 5. The ordinate scale is in arbitrary units.

shown also by the EOG). As a result, the second saccade and the associated lambda responses showed a greater dispersion over time. Factor 2 scores showed a significant electrode position effect ( F ( 5 , 96)--10.32, P < 0 . 0 0 1 ) , the greatest positive value appearing at the occipital site. Factor 3 showed a strong loading in the period

Late lambda components The main goal of our investigation was to describe the behavior of the late positive component(s) of the lambda response in a typical cognitive paradigm. In a guessing task we found associated with the lambda responses a late positive component and a subsequent slow wave which were similar to those typically reported in (traditional) ERP studies in which the eyes were stationary. This conclusion is supported by the following results. In the peak amplitude measures, in contrast to the control condition, the P300 showed a significant increase in the guessing task and was greatest over the parietal region. Furthermore, the PCVA of lambda responses revealed a factor (factor 1) which gave the strongest loading in the latency range of the P300. Also, the topography of its scores, showing a parietal maximum, was in accord with the PCVA results of other ERP studies as well (Simson et al. 1976; K.C. Squires et al. 1977; Friedman et al. 1980; Ruchkin et al. 1980). These authors have also described in the PCVA of ERPs a further factor identified with the slow wave. With respect to its loading pattern a quite similar factor emerged in the PCVA of the lambda responses in our study. At the same time the topography of the factor 2 scores differed somewhat from earlier observations in that the parietal and central scores appeared closer to each other. However, this finding is in accord with the results reported by Ruchkin et al. (1980) and seems to characterize the slow wave topography in guessing tasks. It is also worth mentioning that the loadings of this SW factor began to rise early: at about 180 msec after saccade onset. A similar observation in ERPs has been recently reported by N~i~it~inen et

660

al. (1982). In detection conditions in about half of their records these authors found a slow parietal positive shift starting between 100 and 200 msec after stimulus onset, while in the rest of the ERPs this shift began at about 300 sec following onset. The possible influence of arousal and motivational state on late lambda components We have to consider the possibility that the observed changes in late components could have arisen as a result of differences between the two conditions in arousal or motivational state. We tested this possibility in two ways: first, by comparing in the two conditions the lambda responses associated with the saccades to the middle light stimulus; and second, by comparing the respective CNVs recorded in the two conditions. We found that the lambda responses accompanying the saccades to the middle light were not markedly different in the two conditions (see the first portion of the CNV averages in Fig. 4). Although a small increase in the later positivity can be observed in the parieto-occipital leads during guessing, this change, in itself, can hardly account for the prominent late positive component associated with the saccades to the target lights. The possible dependence of P300 on the preceding CNV has been a problem discussed repeatedly in ERP research. Earlier it was suggested (e.g., Wilkinson and Lee 1972) that the P300 indicates the resolution of the CNV, but several recent studies provided evidence in favor of the independence of the two phenomena (Donchin et al. 1975; Courchesne et al. 1975; Desmedt and Debecker 1979; Ruchkin et al. 1980; McCallum and Curry 1981). The problem seems to be especially relevant in the case of lambda responses where the initiation of an eye movement involves preparation for information pickup which might lead to a closer link between CNV-like anticipation processes and those events indexed by the P300. Although in our records there appeared slight differences in the magnitude of the CNVs between the two conditions, nevertheless, it seems unlikely that these small differences could satisfactorily account for the robust late positivity of the lambda responses elicited by the target stimuli. Such an interpretation is also supported by the results of applying

M. MARTON ET AL.

PCVA to the CNV averages, in which separate factors were shown to be responsible for the CNV and P300 component. Early and middle lambda components (the N45 and the P176) The P176 component of the lambda response was significantly greater in our guessing task than in the control condition. The P176 measured from saccade onset (or the P100 component measured from saccade offset) can be identified with the classical 'lambda wave' (Kurtzberg and Vaughan 1977; Yagi 1979; Scott et al. 1981). Yagi (1982) found the amplitude of this component to be greater during a visual search task than during simple inspection of the same stimulus field, showing the influence of the greater attentional requirements of searching upon this component. It seems likely that the significant increase of the P176 component in our guessing condition was also due to the increased mobilization of attentional processes. There was a further stable difference between the lambda responses in the two conditions. As Figs. 1 and 4 show, in the guessing task the first prominent negative component, the N45, had a greater amplitude in the lambda responses for both the middle and the target stimuli. This component might be sensitive to both stimulus and attentional factors. In an earlier experiment we demonstrated the existence of such a sensitivity to stimulus factors (Marton et al. 1983). In the present experiment, however, stimulus configuration was held constant across the two conditions; therefore it seems likely that in the guessing condition subjects might have paid more attention to the upcoming stimuli than they did in the control condition. Thus it seems plausible to suggest that the increase in the N45 amplitude in the guessing condition reflects the sensitivity of this component to attentional processes. In sum, the results of the present experiment demonstrate that the study of cortical responses time-locked to saccades offers, besides the study of evoked potentials, an important additional alternative for investigating visual information processing. Our results provide support for the hypothesis that the late components of the lambda response

LAMBDA RESPONSE AND GUESSING

reflect the effect of information processing (e.g., 'hypothesis updating') similarly to the way of the late components of traditional ERPs.

Summary The late positive components of lambda responses were studied in a guessing task modified in such a way that subjects had to perform a saccadic eye movement in order to perceive taskrelevant information. Responses from 6 scalp areas were investigated in 9 subjects and in 2 conditions: control and guessing. In both conditions subjects performed two consecutive saccades in a given trial: the first to the middle light, the second to the target area. In the control condition the subjects knew beforehand what the 'target' would be. In the guessing condition they had to make a guess before each trial as to which of the three target stimuli would appear. The target stimuli occurred with unequal probability and were presented in a randomized sequence. Unlike the control condition, the guessing task led to the appearance of a late positive component in the lambda response. Similar to traditional ERP findings, this late positivity showed an amplitude maximum at the parietal area and a peak latency at 375 msec from saccade onset. Furthermore, Principal Component-Varimax Analysis (PCVA) of the lambda responses revealed a first factor giving the strongest loading in the latency range of the P300, and a second factor which was identified as the slow wave. These factors are quite similar to the factors found in the PCVA of ERPs. Our results suggest that the late components of lambda responses reflect the effects of information processing in cognitive tasks similarly to the way the late positive components of ERPs do.

R6sum6 Composantes tardives des potentiels ckrObraux liOs aux saccades, au cours de t&ches de prddiction

Les composantes positives tardives des r6ponses lambda ont 6t6 analys6es dans une tglche de

661

pr6diction en sorte de forcer les sujets a effectuer un mouvement oculaire saccadique pour percevoir l'information li6e ~ la consigne. Les r6ponses de 6 aires du scalp ont 6t6 &udi6es chez 9 sujets dans 2 conditions: t6moin et estimation. Dans les deux conditions, les sujets effectuaient deux saccades successives dans un essai donn& la premi6re vers une lumi6re centrale, la seconde vers la cible. En situation t6moin, les sujets savaient d'avance ce que serait la cible. En situation de pr6diction, ils avaient h deviner, avant chaque essai, lequel des 3 stimulus cibles apparaitrait. Ces stimulus cibles 6taient donn6s selon des probabilit6s in6gales, et pr6sent6s al~atoirement. Contrairement h la situation t6moin, la tfiche de pr6diction produisait une composante positive tardive dans la r6ponse lambda. C o m m e dans les r6ponses 6voqu6es traditionnelles, cette positivit6 tardive pr6sentait une amplitude maximale en pari6tal, avec une latence de 375 msec h partir du d6but de la saccade. De son c6t6, l'analyse en composantes principales (dite VARIMAX), de ces r6ponses lambda, a mis en 6vidence un premier facteur, avec pond6ration maximale dans le domaine des latences de la P300 et un second facteur, identifi6 com/ne l'onde lente. Ces facteurs sont similaires ceux trouv6s dans les potentiels 6voqu6s avec la m~me m6thode d'analyse. Nos r6sultats sugg~rent que les composantes tardives de la r6ponse iambda traduisent les effets du traitement de l'information dans une t~che cognitive, tout comme les composantes positives tardives de la r6ponse 6voqu6e. The authors would like to thank Gy. Gergely and R. Jiirgens for their advice and help during preparation of the manuscript.

References Cooper, R., McCallum, W.C., Newton, P., Papakostopoulos, D., Pocock, P.V. and Warren, W.J. Cortical potentials associated with the detection of visual events. Science, 1977, 196: 74-77. Cooper, R., Pocock, P.V., McCallum, W.C. and Papakostopoulos, D. Potentials associated with the detection of infrequent events in a visual display. In: D.A. Otto (Ed.), Multidisciplinary Perspectives in Event-Related Brain Potential Research, EPA-600/9-77-043. U.S.G.P.O., Washington, D.C., 1978: 189-193.

662 Courchesne, E., Hillyard, S.A. and Galambos, R. Stimulus novelty, task relevance and the visual evoked potential in man. Electroenceph. clin. Neurophysiol., 1975, 39: 131-143. Desmedt, J.E. and Debecker, J. Wave form and neural mechanism of the decision P350 elicited without pre-stimulus CNV or readiness potential in random sequences of nearthreshold auditory clicks and finger stimuli. Electroenceph. olin. Neurophysiol., 1979, 47: 648-670. Donchin, E. Brain electrical correlates of pattern recognition. In: G.F. Inbar (Ed.), Signal Analysis and Pattern Recognition in Biomedical Engineering. Wiley, New York, 1975: 199-218. Donchin, E. Event-related brain potentials: a tool in the study of human information processing. In: H. Begleiter (Ed.), Evoked Brain Potentials and Behaviour. Plenum Press, New York, 1979: 13-88. Donchin, E. and Cohen, L. Averaged evoked potentials and intramodality selective attention. Electroenceph. clin. Neurophysiol., 1967, 22: 537-546. Donchin, E., Tueting, P., Ritter, W., Kutas, M. and Heffley, E., On the independence of the CNV and the P300 components of the human averaged potential. Electroenceph. clin. Neurophysiol., 1975, 38: 449-461. Friedman, D., Hakerem, G., Sutton, S. and Fleiss, JIL. Effect of stimulus uncertainty on the pupillary dilation response and the vertex evoked potential. Electroenceph. clin. Neurophysiol., 1973, 34: 475-484. Friedman, D., Vaughan, Jr., H.G. and Erlenmeyer-Kimling, L. Multiple positive ERP components in visual discrimination tasks. In: H.H. Kornhuber and L. Deecke (Eds.), Motivation, Motor and Sensory Processes of the Brain: Electrical Potentials, Behaviour and Clinical Use. Elsevier/NorthHolland, Amsterdam, 1980:117-122. Gaarder. K.R. Eye Movement, Vision and Behaviour. Wiley, New York, 1975. Hillyard, S.A. and Picton, T.W. Event-related brain potentials and selective information processing in man. In: J. Desmedt (Ed.), Progress in Clinical Neurophysiology. Vol. 6. Cognitive Components in Cerebral Event-Related Potentials and Selective Attention. Karger, Basel, 1978: 1-50. Hillyard, S.A,, Squires, K.C., Bauer, J.W. and Lindsay, P.H. Evoked potential correlates of auditory signal detection. Science, 1971, 172: 1357-1360, Johnson, Jr., R. P300: a model of the variables controlling its amplitude. In: R. Karrer, J. Cohen and P. Tueting (Eds.), Cognitive Processing and Brain Activity. New York Academy of Sciences, New York, 1983: in press. Kurtzberg, D. and Vaughan, Jr., H.G. Electrophysiological observations on the visuomotor system and visual neuro~ sensorium. In: J. Desmedt (Ed.), Visual Evoked Potentials in Man: New Developments. Clarendon Press, Oxford, 1977: 314-331. Kurtzberg, D, and Vaughan, Jr., H.G. Maturation and task specificity of cortical potentials associated with visual scanning. In: D. Lehmann and E. Callaway (Eds.), Human Evoked Potentials: Applications and Problems. Plenum Press, New York, 1979: 185-199.

M. MARTON ET AL. Marton, M., Szirtes, J. and Donauer, N. A comparative study of averaged lambda potentials in man and monkey. In: R. Sinz and M. Rosenzweig (Eds.), Psychophysiology 1980. Fischer, ,lena, and Elsevier, Amsterdam, 1983: 379-389. McCallum, W.C. and Curry, S.H. Late slow wave components of auditory evoked potentials: their cognitive significance and interaction. Electroenceph. clin. Neurophysiol., 1981, 51: 123-137. N~igtt~inen, R., Simpson, M. and Loveless, N.E. Stimulus deviance and evoked potentials. Biol. Psychol., 1982, 14: 53-98. Neisser, U. Cognition and Reality. Freeman and Co., San Francisco, Calif., 1976. Nie, N.H., Hall, C.H., Jenkins, J.G., Stenbrenner, K. and Bent, D.H. Statistical Package for the Social Sciences. McGrawHill, New York, 1972. Parasuraman, R., Richter, F. and Beatty, J. Detection and recognition: concurrent processes in perception. Percept. Psychophys., 1982, 31: 1-12. Pritchard, W.S. Psychophysiology of P300. Psychol. Bull., 1981, 89: 506-540. Ritter, W. Latency of event-related potentials and reaction time. In: D.A. Otto (Ed.), Multidisciplinary Perspectives in Event-Related Brain Potential Research, EPA-600/9-77043. U.S.G.P.O., Washington, D.C., 1978:173 174. Ritter, W. and Vaughan, Jr., H.G. Averaged evoked responses in vigilance and discrimination: a reassessment. Science, 1969, 164: 326-328. Roth, W.T., Ford, J.M. and Koppel, B.S. Long-latency evoked potentials and reaction time. Psychophysiology, 1978, 15: 17 23. Ruchkin, D.S., Sutton, S. and Tueting, P. Emitted and evoked P300 potentials and variation in stimulus probability. Psychophysiology, 1975, 12: 591-595. Ruchkin, D,S., Sutton, S. and Stega, M. Emitted P300 and sk)w wave event-related potentials in guessing and detection task. Electroenceph. clin. Neurophysiol., 1980, 46: I - 14. Scott, D.F., Moffett, A. and Bickford, R.G. Comparison of two types of visual evoked potentials: pattern reversal and eye movement (lambda). Electroenceph. clin. Neurophysiol., 1981, 52: 102-104. Simson, R., Vaughan, Jr., H.G. and Ritter, W. The scalp topography of potentials associated with missing visual or auditory stimuli. Electroenceph. clin. Neurophysiol., 1976, 40: 33-42. Squires, K.C., Donchin, E., Herning, R.i. and McCarthy, G. On the influence of task relevance and stimulus probability on event-related potential components. Electroenceph. clin. Neurophysiol., 1977, 42: 1-14. Squires, N.K., Squires, K.C. and Hillyard, S.A. Two varieties of long-latency positive waves evoked by unpredictable auditory stimuli in man. Electroenceph. olin. Neurophysiol., 1975, 38: 387-401. Squires, N.K., Donchin, E., Squires, K.C. and Grossberg, S. Bisensory stimulation: inferring decision-related processes from the P300 component. J. exp. Psychol. hum. Percept., 1977, 3: 299-315. Sutton, S., Braren, M., Zubin, J. and John, E.R. Evoked

LAMBDA RESPONSE AND GUESSING potential correlates of stimulus uncertainty. Science, 1965, 150: 1187-1188. Sutton, S., Tueting, P., Hammer, M. and Hakerem, G. Evoked potentials and feedback. In: D.A. Otto (Ed.), Multidisciplinary Perspectives in Event-Related Brain Potential Research, EPA-600/9-77-043. U.S.G.P.O., Washington, D.C., 1978:184 188. Szirtes, J., Marton, M. and Breuer, P. Lambda potential in the rhesus monkey. Physiol. Psychol., 1982, 10: 261-266. Tueting, R., Sutton. S. and Zubin, J. Quantitative evoked

663 potential correlates of the probability of events. Psychophysiology, 1970, 7: 385-394. Wilkinson, R.T. and Lee, M.V. Auditory evoked potentials and selective attention. Electroenceph. clin. Neurophysiol., 1972, 33: 411-418. Yagi, A. Saccade size and lambda complex in man. Physiol. Psychol., 1979, 7: 370-376. Yagi, A. Visual signal detection and lambda responses. Electroenceph, clin. Neurophysiol., 1981, 52: 604-610. Yagi, A. Lambda response as an index of visual perception research. Jap. psychoL Res., 1982, 24:106 110.